Development of Soil and Litter Arthropod Assemblages in Rainforest Restoration

Author Nakamura, Akihiro

Published 2008

Thesis Type Thesis (PhD Doctorate)

School Griffith School of Environment

DOI https://doi.org/10.25904/1912/2751

Copyright Statement The author owns the copyright in this thesis, unless stated otherwise.

Downloaded from http://hdl.handle.net/10072/367553

Griffith Research Online https://research-repository.griffith.edu.au

DEVELOPMENT OF SOIL AND LITTER ARTHROPOD ASSEMBLAGES IN RAINFOREST RESTORATION

Akihiro Nakamura

2007

DEVELOPMENT OF SOIL AND LITTER ARTHROPOD ASSEMBLAGES IN RAINFOREST RESTORATION

Akihiro Nakamura B.Sc. (Hons)

Centre for Innovative Conservation Strategies and Griffith School of Environment Faculty of Science, Environment, Engineering and Technology Griffith University

Supervisors Dr Carla P. Catterall Prof Roger L. Kitching Dr Alan P. N. House

Submitted in fulfilment of the requirements of the degree of Doctor of Philosophy September 2007

SYNOPSIS

Broadscale forest clearance is a major human-induced disturbance with devastating consequences for biodiversity. With a rise in public awareness of biodiversity, the last few decades have seen an increasing number of reforestation activities aimed at recreating natural habitat. To date, research on the outcomes of reforestation for biodiversity have mostly focused upon the recovery of vegetation structure and composition, with relatively little attention being paid to the recovery of fauna, especially invertebrates.

Arthropods inhabiting soil and leaf litter constitute a considerable proportion of the biomass and diversity of a given faunal community, and exhibit strong associations with ecological functioning (i.e. soil formation, decomposition). The recovery of soil and litter arthropods in restored landscapes is therefore important, not only for the recovery of biodiversity but also for the re-development of a functioning ecosystem, a goal of most types of reforestation programs. However, we know little about the patterns of development of arthropod assemblages in reforested landscapes or the mechanisms underpinning any such patterns.

The broad objective of this study was to investigate the effects of selected factors on the colonisation patterns of restored rainforest patches by soil and leaf litter arthropods. Criteria for selecting factors for study included their potential influence on the development of arthropod assemblages and their potential for manipulation by restoration practitioners. The study was conducted on the Blackall Range near Maleny, a mid-elevation (250-530 m a.s.l.) basaltic plateau in subtropical eastern Australia (26°S, 152°E). The plateau supported subtropical rainforest until European settlement in the 19th century, when most of the rainforest was cleared for pasture.

The factors selected for study, in relation to the colonisation of restored habitat patches by soil and litter arthropods, were as follows:

1. the isolation of restored habitat patches from remnant forest;

2. the efficacy of inoculation (re-introduction of rainforest soil and litter arthropods) to restored habitat patches;

i 3. the quality and quantity of substrate (i.e. mulch) used during the initial stages of rainforest restoration;

4. the degree of shading and depth of substrate, and their interaction; and,

5. the short- and longer-term impacts of glyphosate herbicide on arthropod assemblages.

To test explicitly the effects of these factors on arthropod colonisation of restored habitat patches, an experimental approach was adopted in this study. The first four factors were addressed by means of a manipulative field experiment. Small-scale habitat patches were created by adding sterilised mulch to an area previously treated with glyphosate herbicide, and covered with shadecloth, to simulate various conditions of forest restoration which may be experienced by colonising arthropods. In order to test for the impacts of a glyphosate herbicide on rainforest arthropods, I carried out a separate field experiment in which experimental patches were established within remnant rainforests. The experimental approach adopted in this study allowed for the construction of replicated units, while controlling for extraneous factors (e.g. heterogeneity of litter composition, habitat area, age of restoration), to enable robust examination of the effects of selected factors.

To monitor assemblage composition, arthropods were collected using two methods: pitfall traps and extraction from litter. Responses of arthropods were analysed at two main levels of taxonomic resolution: ‘coarse’ arthropods (all arthropods sorted to Order/Class) and ants, identified to species.

Before the field experiments, a survey was carried out to collect reference information on the distribution of soil and litter arthropods in remnant rainforests (undisturbed reference sites) and cleared pasture (disturbed reference sites) in the study region. Regardless of the sampling method (pitfall or litter extraction) or taxonomic resolution employed (coarse arthropods or ant species), the composition of arthropod assemblages clearly differed between rainforest and pasture. The information obtained from this survey generated potential bio-indicators of forested and cleared habitats, assisting interpretation of the data obtained from the field experiments.

The effects of habitat isolation and inoculation were tested using ‘restored’ habitat patches which were established within cleared pasture at increasing distances (0, 15, 100 ii and ca. 400 m) from the edges of rainforest remnants. After nine months, rainforest-dependent taxa were found to have only colonised the habitat patches closely adjacent to rainforest remnants. Attempts to increase the extent of arthropod establishment by inoculation were unsuccessful: the majority of rainforest arthropods from the raw inoculum failed to persist within the isolated plots. The results indicated that many forest-dependent soil and litter arthropods may have a limited capacity to colonise isolated restoration sites, and/ or small experimental plots in the short term. Inference from the experiment was potentially limited by the relatively small temporal and spatial scales of restoration treatments. Avoiding these limitations in future research may require controlled and replicated efforts in experimental restoration over larger areas, based on collaborations between researchers and practitioners.

To test the effects of quality and quantity of substrate on arthropod colonisation, habitat patches were established by adding either sterilised hay (a conventionally used mulching material in restoration projects) or woodchip mulch (a structurally more complex alternative), each at two depths (3-5 cm, 10-15 cm). Habitat patches were positioned within pasture adjacent to the edges of rainforest remnants to minimise the effect of isolation, and were all unshaded to create conditions similar to those during the initial stages of rainforest restoration. Despite its simple composition, hay performed better than woodchips in facilitating colonisation by arthropods characteristic of rainforest. However, neither hay nor woodchip mulch inhibited arthropods invading from the surrounding pasture. Shallow hay was favoured by ants characteristic of rainforest, but other groups of arthropods (e.g. Coleoptera, Isopoda) were associated with deep hay (10-15 cm). The optimum amount of hay may therefore vary among different groups of arthropods.

The effects of shading and mulch depth, and potential interactions between them, were tested using habitat patches created with varying degrees of shading (none, 50% or 90% shading) and two depths of woodchip mulch (3-5 cm or 10-15 cm deep). The presence of shading, at both 50% and 90%, encouraged colonisation of habitat patches by arthropods characteristic of rainforest. However, only the more complete shading treatment (90%) inhibited re-invasion of restored patches by arthropods from the surrounding pasture habitat. Effects of mulch depth were significant only for rainforest-associated ant species which responded positively to shallow mulch within shaded plots. These results suggest that moderate levels of canopy closure, as produced by tree spacings typical of timber plantations, may be sufficient to facilitate colonisation of reforested land by rainforest iii arthropods. However, greater canopy shading (90%) is likely to be needed to inhibit re-invasion of arthropods from surrounding pasture habitat. Using deeper woodchip mulch does not necessarily create more suitable conditions for rainforest arthropods or offset the deleterious effects of the lower levels of shading.

Paired herbicide-treated and control plots were created within rainforest remnants to test the short- (approximately three days) and long-term (approximately three months) impacts of herbicide application on soil and litter arthropod assemblages on the floor of the remnant rainforests. The results found no deleterious effects of glyphosate herbicide formulated as Roundup® Biactive™ on rainforest soil and litter arthropods; hence, this herbicide appears suitable for the control of unwanted plants in rainforest restoration projects, from the perspective of arthropod biodiversity.

The outcomes of this study also have a number of important implications for the monitoring of the development of soil and litter arthropod assemblages in restored rainforests. First, pitfall traps (a commonly used sampling technique) can provide sufficient information on the state of arthropod assemblages in the context of subtropical rainforest restoration projects, although samples collected by this method alone do not necessarily represent the whole suite of soil and litter arthropods (e.g. cryptic arthropods that live in soil and litter). Second, a combination of higher-taxon sorting of all arthropods, together with species-level sorting of a significant major taxon (ants), provides a feasible compromise between comprehensiveness and detail in monitoring responses of arthropods. Third, ‘composite habitat indices’, such as those developed in this study to quantify the extent to which a site resembles rainforest or pasture in terms of its arthropod assemblage, can help alleviate problems associated with the patchy distribution of arthropod taxa in monitoring samples.

The experimental approach adopted in this study provided information that would otherwise have been limited by post-hoc empirical studies alone. This study’s results demonstrated that the selected aspects of different restoration techniques and management affect the colonisation of soil and litter arthropod assemblages in rainforest restoration of old fields.

iv

STATEMENT OF ORIGINALITY

I certify that this thesis is my original work and has not previously been submitted for a degree or diploma in any university. To the best of my knowledge and belief, the thesis contains no material previously published or written by another person except where due reference is made in the thesis itself.

______Akihiro Nakamura

v TABLE OF CONTENTS

SYNOPSIS ...... i STATEMENT OF ORIGINALITY ...... v TABLE OF CONTENTS ...... vi LIST OF FIGURES...... viii LIST OF TABLES ...... xi LIST OF APPENDICES ...... xv ACKNOWLEDGEMENTS ...... xvi COLOUR PLATES ...... xix PREFACE...... xxiii

1 GENERAL INTRODUCTION ...... 1 1.1 Restoration of cleared landscapes...... 1 1.2 Restoration of cleared rainforest in Australia...... 3 1.3 Monitoring soil and litter arthropods in restored habitats ...... 4 1.3.1 Reasons for monitoring soil and litter arthropods ...... 4 1.3.2 Approaches to cost-effective monitoring with soil and litter arthropods ...... 7 1.3.3 Studies of soil and litter arthropods in actively-restored lands ...... 12 1.4 Aims and thesis structure...... 23

2 OVERVIEW OF METHODOLOGY...... 27 2.1 Study area...... 27 2.2 Study design...... 29 2.2.1 Components of the study ...... 29 2.2.2 Manipulative field experiment...... 29 2.2.3 Pilot studies ...... 34 2.3 Sampling methodology...... 35

3 BASELINE SURVEY...... 37 3.1 Introduction ...... 37 3.2 Methods ...... 38 3.3 Results ...... 42 3.4 Discussion...... 48

4 EFFECTS OF ISOLATION AND INOCULATION...... 53 4.1 Introduction ...... 53 4.2 Methods ...... 55 4.3 Results ...... 60 4.3.1 Isolation ...... 61 4.3.2 Inoculation...... 70 4.4 Discussion...... 72

5 EFFECTS OF MULCH QUALITY AND DEPTH...... 77 5.1 Introduction ...... 77 5.2 Methods ...... 79 5.3 Results ...... 84 5.4 Discussion...... 98

vi 6 EFFECTS OF SHADING AND MULCH DEPTH ...... 103 6.1 Introduction ...... 103 6.2 Methods ...... 105 6.3 Results ...... 110 6.4 Discussion...... 123

7 EFFECTS OF HERBICIDE APPLICATION...... 127 7.1 Introduction ...... 127 7.2 Methods ...... 129 7.3 Results ...... 132 7.4 Discussion...... 138

8 GENERAL DISCUSSION...... 141 8.1 Summary and key findings of the data chapters (Chapters 3-7) ...... 141 8.2 Experimental approaches in restoration studies ...... 145 8.3 Use of reference information in restoration studies...... 147 8.4 Choice of target taxa, measurements and methods for monitoring the state of soil and litter arthropods in restored rainforests...... 149 8.5 Introduced species in restored rainforests ...... 153 8.6 Recommendations for restoration practice...... 153

REFERENCES ...... 157

APPENDIX 1 ...... 191 APPENDIX 2 ...... 197 APPENDIX 3 ...... 199 APPENDIX 4 ...... 205

vii LIST OF FIGURES

Figure 1.1 Conceptual flow chart of the thesis data-chapters, explaining aims and interrelationships with other data-chapters (indicated by arrows)...... 25 Figure 2.1 A map of the study area, showing 12 paired sites used in the baseline survey (Chapter 3) and five experimental sites used in Chapters 4-7. Each experimental site comprised a combination of one main (E1-5) and one ‘distant’ sub-site (D1-5) (see text for more details)...... 27 Figure 2.2 Schematic diagram of one of the field experimental sites (distances not to scale), showing criteria used when positioning experimental plots. ‘Deep’ indicates sterilised ‘woodchip’ or hay mulch placed at a depth of 10-15 cm, and ‘shallow’ is 3-5 cm. The plots 1-4 and 14 contained deep woodchip mulch. The nine plots along the edge were located randomly. Plot numbers in parentheses corresponds to those in Table 2.1. See also Plates 3, 5-8...... 32 Figure 2.3 Layout of an experimental plot (not to scale). When present, shadecloth covered the area within the solid line...... 33 Figure 3.1 NMDS ordination of rainforest and pasture sites based on (a) coarse arthropods, (b) ant genera, (c) ant species and (d) ant functional groups. Points sampled using pitfall traps and litter extraction are shown separately. Note that two sites with low grazing intensity are indicated by crossed boxes...... 43 Figure 4.1 NMDS ordination of experimental plots at various distances from the rainforest edge, and rainforest and pasture reference habitats. Assemblages sampled using different sampling methods (pitfall traps, litter extraction) are shown separately...... 62 Figure 4.2 Mean taxon richness and composite rainforest and pasture indices of coarse arthropods and ant species across the five different distances (-30 m, 0 m, 15 m, 100 m, 400 m) from rainforest edge. Values for rainforest (RF) and pasture (P) reference habitats (sampled the previous year) are also shown. Open bars, pitfall traps; closed bars, litter extraction...... 65 Figure 4.3 Abundances (mean, SE) of the individual arthropod taxa that showed statistically significant differences across the five different distances (-30 m, 0 m, 15 m, 100 m, 400 m) from the rainforest edge. Values for rainforest (RF) and pasture (P) reference habitats (sampled the previous year) are also shown. Open bars, pitfall traps; closed bars, litter extraction...... 68 Figure 4.4 Abundance scores (mean, SE) of the individual ant species that showed statistically significant differences across the five different distances (-30 m, 0 m, 15 m, 100 m, 400 m) from the rainforest edge. Values for rainforest (RF) and pasture (P) reference habitats (sampled the previous year) are also shown. Open bars, pitfall traps; closed bars, litter extraction...... 69 Figure 4.5 Soil moisture content (mean, SE) across the five different distances (-30 m, 0 m, 15 m, 100 m, 400 m) from rainforest edge, and for rainforest (RF) and pasture (P) reference habitats, across all sites. All samples were taken in 2004...... 70 Figure 4.6 NMDS ordination of experimental plots comparing inoculated and no-inoculation treatments at 400 m from the rainforest edge. Assemblages extracted from the raw inoculum are also shown for litter-extracted samples (b, d)...... 71 viii Figure 5.1 NMDS ordination of the experimental plots and rainforest and pasture site based on (a, b) coarese arthropods and (c, d) ant species. Assemblages sampled by different sampling methods (pitfall traps, litter extraction) are shown separately...... 86 Figure 5.2 Effects of mulch quality and depth on composite rainforest and pasture indices of coarse arthropods (mean index level, SE). Values for pasture and rainforest reference habitats (sampled the previous year) are also shown. The plots included in the analysis of two-factor crossed ANOVA (Table 5.2) were represented by closed bars. Wood = woodchip mulch...... 87 Figure 5.3 Effects of mulch quality and depth on abundances (mean, SE) of pitfall-trapped coarse arthropod taxa that showed statistically significant differences. Values for pasture and rainforest reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars. Wood = woodchip mulch...... 92 Figure 5.4 Effects of mulch quality and depth on abundances (mean, SE) of litter-extracted coarse arthropod taxa that showed statistically significant differences. Values for pasture and rainforest reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars. Wood = woodchip mulch...... 93 Figure 5.5 Effects of mulch quality and depth on composite rainforest and pasture indices of ant species (mean index level, SE). Values for pasture and rainforest reference habitats (sampled the previous year) are also shown. The plots included in the analysis of two-factor crossed ANOVA (Table 5.2) were represented by closed bars. Wood = woodchip mulch...... 94 Figure 5.6 Effects of mulch quality and depth on abundance scales (mean, SE) of selected ant species from pitfall traps. Values for pasture and rainforest reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars. Wood = woodchip mulch...... 95 Figure 5.7 Effects of mulch quality and depth on soil moisture contents (mean, SE) across the experimental plots. Values for pasture and rainforest reference habitats are also shown. All samples were taken in 2004. The plots included in the statistical analysis were represented by closed bars. Wood = woodchip mulch...... 96 Figure 5.8 Temperature fluctuations over five days in late summer (April) recorded at: (a) un-mulched control plots, pasture and rainforest reference habitats, and (b) plots with mulching treatments, with mean values and coefficient of variations (CV). Data from one site only (E3, D3)...... 97 Figure 6.1 Effect of shading and mulch depth on composite rainforest (RF) and pasture (PA) indices of coarse arthropods (mean index level, SE). Values for rainforest and pasture reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars...... 113 Figure 6.2 (2 pages) Effect of shading and mulch depth on abundances (mean, SE) of pitfall-trapped (a-g) and litter-extracted (h-j, next page) coarse arthropod taxa: results only for taxa where the main effects of ANOVA P ≤ 0.10 (see also Table 6.2). Values for rainforest and pasture reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars...... 115

ix Figure 6.3 Effect of shading and mulch depth on composite rainforest (RF) and pasture (PA) indices of ant species (mean index level, SE). Values for rainforest and pasture reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars...... 118 Figure 6.4 Effect of shading and mulch depth on abundance scales of ant species: results only for species where the main effects of ANOVA P ≤ 0.10 (see also Table 6.4). Values for rainforest and pasture reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars.120 Figure 6.5 Average soil moisture contents across the experimental treatments and unshaded, un-mulched controls. Values for rainforest and pasture reference habitats are also shown. All samples were taken in 2004. The plots included in the two-factor ANOVA were represented by closed bars...... 121 Figure 6.6 Temperature fluctuations over five days recorded at: (a) un-mulched, unshaded control plots, rainforest and pasture reference habitats, and (b) plots with different shading and mulch depths, with mean values and coefficient of variations (CV). Temperature was recorded only from one site (E3, D3) during April 2004. Horizontal lines are drawn at 25 C˚...... 122 Figure 7.1 NMDS ordinations based on log-transformed abundances of (a) coarse arthropods and (b) ant species before and after the herbicide application (three days and three months after application) for the control and herbicide-treated plots. NMDS ordinations based on presence/absence data are not shown as they produced similar patterns to those based on log-transformed abundance data...... 133 Figure 7.2 Total abundances, taxon richness, and abundances (mean, SE) of individual arthropod taxa/species before and after the herbicide application for the control (open circle) and treated (closed square) plots. Figure shows only common taxa/species that occurred in all the plots before application, or that showed significant responses to the experimental treatments (P < 0.05) (see also Table 7.3)...... 137

x LIST OF TABLES

Table 1.1 Summary of ant functional groups with some of the major genera found in Australia, North America or Europe (after Andersen 1997b; Gomez et al. 2003). See also Brown (2000) for the ant genera of the world with their associated functional groups...... 13 Table 1.2 The number of studies of actively-restored vegetation which incorporated soil and litter arthropods, carried out in different regions and habitat types. Studies are subdivided into minesite restoration, old field restoration and other types of restoration. Note that the total number of studies from habitat types does not match the actual number of studies summarized as one study often straddled two or more habitat types...... 14 Table 1.3 The number of studies of actively-restored vegetation incorporating groups of soil and litter arthropods at different taxonomic and functional ranks. Note that one study often investigated two or more groups of arthropods with various taxonomic and functional ranks...... 16 Table 1.4 The number of studies of actively-restored vegetation involving soil and litter arthropods that did or did not include reference sites, and the number of spatial replications employed. Reference sites were subdivided into either ‘undisturbed’ (i.e. intact habitats or old secondary-growth forests, generally representing the ‘goals’ of restoration projects), or ‘disturbed’ (habitat representing the status before restoration commenced)...... 18 Table 1.5 The number of studies investigating the effect of various factors on the occurrence of soil and litter arthropods in actively-restored terrestrial habitats. Included factors were investigated by at least two restoration studies. When a given factor was found to influence colonisation patterns of arthropods, it was tallied into either: ‘significant’ when that factor was tested by means of statistical analyses and/or graphical representations; or ‘potential’ when only anecdotal evidence was provided. 21 Table 2.1 Description of experimental plots established at each site (see also Figure 2.2). ‘Woodchip’ or hay mulch was placed at a depth of 10-15 cm (‘deep’), or 3-5 cm (‘shallow’). Shadecloth rated at either 50% or 90% protection from insolation was used to cover some of the experimental plots. See text for more details...... 31 Table 2.2 Sampling methods and taxonomic resolutions of arthropods used in respective chapters (indicated by ‘Y’ symbols)...... 35 Table 3.1 Global R values of each dataset comprising assemblages of rainforest and pasture, using different sampling methods. Note that all P values for Global R were < 0.001...... 44 Table 3.2 Mean abundances and Indicator values (IndVals) of coarse arthropod habitat indicators. Only statistically significant indicators for each sampling method are shown...... 45 Table 3.3 Mean frequency scores and Indicator values (IndVals) of ant genus habitat indicators. Only statistically significant indicators for each sampling method are shown...... 45 Table 3.4 Mean frequency scores and Indicator values (IndVals) of ant species habitat indicators. Only statistically significant indicators for each sampling method are shown...... 47

xi Table 3.5 Mean relative abundances and Indicator values (IndVals) of ant functional group habitat indicators. Only statistically significant indicators for each sampling method are shown...... 48 Table 3.6 Mean index levels and Indicator values (IndVals) of composite indices under each habitat indicator group. Composite indices of ‘specialists’ only (S), ‘increasers’ only (I) and combined ‘specialists’ and ‘increasers’ (S + I) are shown separately. IndVal was not calculated for a group containing less than two component taxa (T1 < 2)...... 49 Table 4.1 ANOSIM global R and pair-wise R values comparing coarse arthropod assemblage compositions between treatments, for each trapping method. Significant P values (< 0.05) are shown in bold...... 63 Table 4.2 ANOSIM global R and pair-wise R values comparing ant species assemblage compositions between treatments, for each trapping method. Significant P values (< 0.05) are shown in bold...... 64 Table 4.3 Summary results of repeated measures ANOVA for (a) coarse arthropods and (b) ant species. The results of individual habitat indicators are shown only when their abundance (or abundance scale) responded significantly to the experimental treatments. Across the habitat boundary, tests compare plots at -30 m, 0 m, and 15 m. Away from habitat boundary, tests compare plots at 15 m, 100 m, and 400 m. Significant P values (< 0.05) are shown in bold...... 66 Table 5.1 Effects of mulch quality (hay, woodchip) and depth (‘shallow’, ‘deep’) on assemblage composition of (a) coarse arthropods, and (b) ant species. F and P values are from non-parametric MANOVA. Df for mulch quality (Quality), mulch depth (Depth), their interaction and site are 1, 1, 1, 4 respectively...... 87 Table 5.2 Effects of mulch quality (hay, woodchip) and depth (‘shallow’, ‘deep’) on levels of composite rainforest and pasture indices for (a) coarse arthropods, and (b) ant species. F and P values are from two-factor crossed ANOVA with randomized complete block design. Df for mulch quality (Quality), mulch depth (Depth), their interaction and site are 1, 1, 1, 4 respectively. Some composite indices were not calculated, as less than 2 component species (T < 2) were found...... 88 Table 5.3 Effects of mulch depth involving the un-mulched control, ‘shallow’ mulch and ‘deep’ mulch treatments, on levels of composite rainforest and pasture indices for (a) coarse arthropods, and (b) ant species. Analysis was carried out separately for woodchip and hay mulch. F and P values are from single-factor ANOVA with randomized complete block design. Df for mulch depth (Depth) and site are 2 and 4 respectively. Some composite indices were not calculated, as less than 2 component species (T < 2) were found...... 89 Table 5.4 Effects of mulch quality (hay, woodchip) and depth (‘shallow’, ‘deep’) on abundances of pitfall-trapped and litter-extracted coarse arthropods. Only taxa that had significant response to the treatments are shown. Their habitat indicator category (rainforest indicator, pasture indicator, ‘generalist’) is shown in parenthesis. F and P values are from two-factor crossed ANOVA with randomized complete block design. Df for mulch quality (Quality), mulch depth (Depth), their interaction and site are 1, 1, 1, 4 respectively...... 91

xii Table 6.1 Effects of shading and mulch depth on assemblage composition and composite rainforest and pasture indices of coarse arthropods. Between-site effects are also shown. ‘Difference’ shows the results of post-hoc permutation (for assemblage composition) or LSD (for composite indices) tests (% levels with different letters are significantly different, P < 0.05). Df for shading, mulch depth, interaction and site are 2, 1, 2, and 4 respectively. Composite habitat indices were not calculated for all ‘specialist’ indicators as they occurred at less than four plots, or were absent altogether...... 112 Table 6.2 Effects of shading and mulch depth on abundances of coarse arthropod taxa: results only for taxa where the main effects of ANOVA P ≤ 0.10. Between-site effects are also shown. ‘Difference’ shows the results of LSD tests (levels with different letters are significantly different; A smaller, B larger, P < 0.05). Df for shading, mulch depth, interaction and site are 2, 1, 2, and 4 respectively...... 114 Table 6.3 Effects of shading and mulch depth on assemblage composition and composite rainforest and pasture indices of ant species. Between-site effects are also shown. ‘Difference’ shows the results of post-hoc permutation (for assemblage composition) or LSD (for composite indices) tests (% levels with different letters are significantly different, P < 0.05). Df for shading, mulch depth, interaction and site are 2, 1, 2, and 4 respectively. Composite indices were not calculated for all litter-extracted ants, as less than 2 component species (T† < 2) were found from each index...... 117 Table 6.4 Effects of shading and mulch depth on abundance scales of ant species: results only for species where the main effects of ANOVA P ≤0.10. Between-site effects are also shown. ‘Difference’ shows the results of LSD tests (levels with different letters are significantly different; A smaller, B larger, P < 0.05). Df for shading, mulch depth, interaction and site are 2, 1, 2, and 4 respectively...... 119 Table 7.1 ANOSIM global R values comparing arthropod assemblage compositions of control versus herbicide-treated plots. Arthropod compositions were compared using all samples, samples collected before application, three days after application and three months after application. Both log-transformed abundance and presence/absence data were used for coarse arthropods and ant species...... 133 Table 7.2 Mean and standard errors of Bray-Curtis distance values that measured dissimilarities of arthropod assemblage compositions before versus after the herbicide application (three days or three months after application) at each site. Differences in the distance values between control and herbicide-treated plots were tested using randomization-based paired-t-test. Both log-transformed and presence/absence data were used for coarse arthropods and ant species...... 134 Table 7.3 (2 pages) Effects of control/treated (control versus treated) and time (before, three days, three months after application) on abundances and taxon richness of (a) coarse arthropods and (b) ant species (next page): results for variables that occurred in at least four samples (N = 30). The effect of the herbicide application was indicated by the interaction of the two factors. Significant values (P < 0.05) are shown bold...... 135 Table 8.1 General patterns in the levels of composite rainforest/pasture indices that were found in factorial analyses of the manipulative field experiment (Chapters 4, 5 and 6). Levels of composite indices under different experimental treatments are represented comparatively by “>”, “<” or “=” symbols. The results are presented only for composite habitat indices that showed statistically significant responses (P < 0.05)...... 143

xiii Table 8.2 Information to aid decisions about monitoring the development of soil and litter arthropod assemblages in rainforest restoration...... 154 Table 8.3 Information to aid decisions in rainforest restoration management, based on the findings of this project...... 156

xiv LIST OF APPENDICES

Appendix 1 Summary of the previous studies of actively-restored terrestrial habitats which incorporated soil and litter arthropods...... 191

Appendix 2 Detailed description of the study sites...... 197

Appendix 3a, b Checklist of (a) coarse arthropod taxa and (b) ant species found in the reference sites, experimental plots and raw inoculum...... 199

Appendix 4a, b Supplementary figures based on the results of the baseline survey (Chapter 3)...... 205

xv ACKNOWLEDGEMENTS

First and foremost, I thank my principal supervisor, Carla Catterall for her support, guidance and insightful criticisms of the draft thesis chapters. She not only made her self available to discuss about issues that arose over the last 5+ years, but also provided me with invaluable inspiration and information. She also gave me an opportunity to join her discussion group meetings where her students and colleagues exchanged inspirational opinions and critical comments on many ecological issues. Without her patience and continuous support, this work would have never been completed. Thank you Carla.

I thank my associate supervisors, Roger Kitching and Alan House for their support, encouragement and critical comments on the draft thesis chapters. Roger provided me with invaluable inspiration and critical comments on the design of the field experiment. Alan gave me lots of helpful comments on study design, data analyses and draft thesis chapters. Alan’s kind support always encouraged me when I was badly stressed.

I also thank my previous supervisor, Heather Proctor, who unfortunately left Griffith University during the first year of my candidature. Without her support and insightful guidance during my honours degree and the early period of PhD candidature, this PhD project would have never been possible. She also coined the acronym for the manipulative field experiment (MaLIRE: Maleny Litter Invertebrate Restoration Experiment), which has been used throughout the duration of my study.

Chris Burwell, my friend and a superb taxonomist, generously provided me with a wealth of taxonomic knowledge and helped me identify ant species whenever I brought a huge number of specimens in ‘pizza boxes’. I am also grateful for his comments on the draft thesis chapters as well as his great sense of humour.

I am also grateful to my colleague and friend, John Kanowski, for his critical comments on my earlier study designs and insightful and clinical comments on draft thesis chapters.

I am indebted to Cas Vanderwoude of the Queensland Department of Primary Industries Fire Ant Research Centre. Cas generously provided the steaming facilities which I was struggling to find. Without his support, the manipulative field experiment (the main part of this PhD project) would have stalled. I am also grateful to Steve Henry and his staff at Steam Services for their help. I thank Mervyn Carey who generously turned his truck into xvi a steaming container and helped me handle massive loads of ‘cooked’ mulch. Woodchip mulch and hay were supplied by Smokey Flats Timber Cutting Group (Wacol, Queensland) and The Hay Man Nursery (Peachester, Queensland). I also acknowledge Damien Clark of Metalcorp Steel who gave me a massive discount on the materials to construct the field experiment.

A special thankyou to Alan Andersen for providing both taxonomic and ecological knowledge of ants. Big thanks also to: Marti Anderson for her advice on multivariate analyses, Ben Harms for identification of the soil types, Bill McDonald for his help on rainforest classification, and to Deanna Tomerini for her generous provision of rainfall data.

This study would not have been possible without the support of the Maleny landholders, Rob and Anne Cork, Colin Cork, Paul Cussack, Ray Dailey, Richard Dent and Jill Morris, John Evans, Jeff Green, Karen and Ed Lawley, Neil MacLeod, Gary Martin, Graham Newton, Graham Nott, David Rowland, John Thirnbeck, Mal and Merg Thompson, Phil and Kate Vickers, who generously lent parts of their properties for the construction of the experimental plots, and allowed full access for over 2 years of the field surveys and experiments. Gary, I know you are a good man even though you tried to threaten me with your shotgun. I thank Rob and Anne Cork for their warm hospitality when I stayed overnight at their home after I got my vehicle stuck at the bottom of the gully. Rob and his brother recovered the vehicle next morning and I suddenly became a famous “Oh, you are that tall Japanese guy…” wherever I went in Maleny. I regret that I also got bogged in the paddocks of Phil and Kate Vickers, and Richard Dent and Jill Morris, and I am very grateful for their generous help and support in the recovery of my vehicle.

I owe a great deal of gratitude to the members of Barung Landcare in Maleny for the wealth of their local knowledge. Special thanks to Mim Coulstock, Josi Marriott and Marc Russel for introducing me to the landholders in Maleny.

I just cannot thank my friends, Michelle Baker, John Holt, Minako Sato, Junko Sato, Yuko Watanabe, Ray Wong enough for their laboratory and (very laborious) field assistance. Thankyou also to Fred Beaulieu for his advice on my draft proposal and design of the field experiment. His sense of humour was just so unique and inspirational!

xvii I was fortunate that I shared lots of inspiration and discoveries with my friends in the Wildlife Ecology Lab and the Kitching Lab where I spent most of the time during the course of my studies. Special thanks to late Scott Piper for his invaluable advice and assistance on statistics. His brilliant ideas always inspired me to develop my intellectual thinking. I miss you man. Thanks to Sarah Boulter, Rod Eastwood and John Holt for their kind assistance and advice.

I was lucky that I shared the ‘triangular’ office room with Darren Bito, Pete Grimbacher, Terry Reis and Tang Yong with whom I shared lots of (often naughty) laughs and inspiration. I thank them all for their friendship and encouragement.

I thank many members of the Griffith School of Environment. Special thanks go to Dave Henstock, Bruce Mudway and Don Dennis for logistical assistance and their great sense of humour; Petney Dickson and Liz Wright for administrative assistance (they are definitely NOT those of the pathetic bureaucrats infesting this university!); and Carmel Wild for proof reading some of my draft thesis chapters. I also wish to thank Clyde Wild for his comments on my PhD proposal.

Generous research funding was provided through the Queensland Government’s Growing Smart State Program. Special thanks to Ann Fisher (coordinator of the program) for all her administrative assistance and advice. Harry Hines, a mentor of this program, gave me invaluable assistance and feedback on the draft thesis chapters. Thank you Harry! Other research funding was provided by the Rainforest Cooperative Research Centre. My stipend was provided by Griffith University Postgraduate Research Scholarship and Griffith University’s Completion Assistance Postgraduate Research Scholarship. Tuition fees were supported by Partial Tuition Fee Scholarship from the Griffith School of Environment.

Finally, and most of all, I wish to thank my partner, Tan Kemkum, for her understanding, patience, support and love throughout the course of my study. I am definitely indebted to Tan’s patience when I filled our home with the smell of ethanol while I was sorting samples in our bedroom. My mother, Noriko Nakamura, visited from Japan and stayed here during the first year of my candidature. I thank her for her love, encouragement, and great cooking! She was also one of the best volunteers in the field.

「タンちゃん、お母さん、ボランティアのみんな、本当に有難う!」 xviii COLOUR PLATES

Plate 1 Left: rainforest (complex notophyll vine forest) in south-east Queensland (Photo: Yong Tang). Top right: pasture typically seen on the Maleny plateau. Bottom right: rainforest-pasture edge at one of the experimental sites (E1).

a c

b d

Plate 2 Ant species typically found in rainforest (a, Discothyrea QM1; b, Prolasius QM2) and pasture (c, Rhytidoponera metallica; d, Paratrechina QM2) on the Maleny plateau (see Appendix 3b for the checklist of ant species) (Photos: David Fleming).

xix

Plate 3 Layout of the experimental plots (see also Chapter 2). Top: plots with no shadecloth constructed along the rainforest edge (cf. Figure 2.2). The photo was taken before mulch was added. Middle: plots covered with 90% shadecloth. Bottom left: a plot positioned at 100 m from the forest edge (the photo was taken from the forest edge). Bottom right: paired plots were established at ca. 400 m from the forest edge (position of the other plot of the pair is indicated by an arrow). xx

Plate 4 Mulch sterilisation and distribution (Chapter 2). Left: mulch was steam-treated in the open-top steel container of a landscaping dump truck. Top right: a tray-back utility vehicle was used to distribute the treated mulch to each of the experimental plots outside rainforests. Bottom right: a 90 L bin was used to distribute the treated mulch to the rainforest control plots (the ‘-30 m’ treatment).

a b

c

Plate 5 Control plots. Top: rainforest control plots (the ‘-30 m’ treatment) in (a) complex notophyll vine forest, and (b) notophyll feather palm vine forest. Note that the rainforest control plots were covered with a tent of plastic bird netting to minimise the natural fall of rainforest leaves and twigs into the experimental area. (c): un-mulched, unshaded control at forest edge.

xxi

Plate 6 Experimental plots along forest edges. Left: a plot with 50% shadecloth and ‘deep’ woodchip mulch. Right: a plot with 90% shadecloth and ‘shallow’ woodchip mulch.

Plate 7 Experimental plots with no shadecloth. Left: a plot with ‘deep’ hay mulch. Right: a plot with ‘shallow’ woodchip mulch. The left photo was taken immediately after mulch was added to the plot; the right photo was taken approximately nine months after the plot was constructed. Regrowth of herbaceous plants occurred in unshaded and some of the lightly shaded (50%) plots.

Plate 8 Experimental plots at 15 m (left) and ca. 400 m (right) from the forest edge. Some of the 400 and 100 m plots were reinforced by extra steel posts. xxii PREFACE

Chapters 3 through 7 were prepared in the form of manuscripts for publication in peer-reviewed journals. These chapters have been slightly modified for inclusion within this thesis. I undertook construction of the field experiments and data collection with help from volunteers (see Acknowledgements). Species-level identification of ants was conducted with help from Dr Chris Burwell. With guidance and advice from my supervisors, I carried out statistical analyses and wrote the initial manuscripts. Co-authors (my three supervisors and Dr Chris Burwell) as well as my other colleagues (Dr John Kanowski, Dr Harry Hines and anonymous journal referees) contributed to editing of the manuscripts.

Chapter 3 has been published in the Journal of Insect Conservation and Chapter 4 has been published in Insect Conservation and Diversity.

Papers published during the course of this research program

Refereed journal articles

Chapter 3: Nakamura, A., Catterall, C.P., House, A.P.N., Kitching, R.L., & Burwell, C.J. (2007) The use of ants and other soil and litter arthropods as bio-indicators of the impacts of rainforest clearing and subsequent land use. Journal of Insect Conservation, 11, 177-186.

Chapter 4: Nakamura, A., Catterall, C.P., Kitching, R.L., House, A.P.N., & Burwell, C.J. (2008) Effects of isolation on the colonisation of restored habitat patches by forest-dependent arthropods of soil and litter. Insect Conservation and Diversity, 1, 9-21.

Conference proceedings

Nakamura, A., Catterall, C.P., Kitching, R.L., & House, A.P.N. (2005) Development of soil and litter arthropod assemblages in rainforest restoration. The International Forestry Review 7: 326.

xxiii xxiv 1 GENERAL INTRODUCTION

1.1 Restoration of cleared landscapes

Despite increasing concerns from both the scientific community and the general public, degradation and destruction of natural habitats has continued at an alarming rate in many regions of the world. The world’s forest cover, for example, has declined by over 13 million hectares every year between 2000 and 2005, mostly through clearance for agricultural or urban use (FAO 2006). Continuing exploitation of the land, in addition to having predicted impacts on climate, is anticipated to cause substantial and irreversible damage to global biodiversity (Sala et al. 2000). Clearing and fragmentation of natural habitats not only causes the loss of many species that depend on specific types of habitats (Watt et al. 1997; Pimm et al. 2001; Jenkins 2003), but also accelerates invasion by exotic species, causing biotic homogenization across continents (Olden 2006). Furthermore, habitat clearance and subsequent land use often degrade ecological functioning such as water retention, soil stability and nutrient cycling, causing devastation to human societies as well as other organisms which live on or nearby the affected land (Gentry & Lopez-Parodi 1980; Pimm et al. 2001; Asner et al. 2004; Foley et al. 2005; Mooney et al. 2005).

With a rise in public awareness of land degradation and biodiversity loss, the last few decades have seen an increasing number of restoration activities, as well as associated scientific studies, aimed at recreating natural habitats (Bennett et al. 2000; Young et al. 2005). Such activities are often termed ‘ecological restoration’: defined by the Society for Ecological Restoration International (2004) as ‘the process of assisting the recovery of an ecosystem that has been degraded, damaged, or destroyed’. To initiate the recovery of an ecosystem, stressors (e.g. mining, agricultural activities) that caused land degradation must first be removed (Brown & Lugo 1994). This is often followed by active human interventions, such as weeding, soil amelioration, seeding and replanting, which may be essential where biotic and abiotic integrities are degraded to the level at which natural recovery is no longer likely (House 1997; Whisenant 1999), or if one wishes to accelerate and guide the process of natural recovery (Lamb et al. 2005).

Recovery of disturbed vegetation can occur with or without active human intervention. Natural regrowth is a common form of terrestrial restoration in which ecosystem recovery

1 takes place without human intervention (Brown & Lugo 1994). This form of recovery is typically seen in ‘old-field’ situations, which occur when cleared land is abandoned following agricultural or pastoral uses (Brown & Lugo 1990; Finegan 1996; Yates & Hobbs 1997). Natural regrowth alone, however, may take a very long time, possibly spanning several decades or even centuries to regain the pre-disturbance state of the habitat (Finegan 1996; Guariguata & Ostertag 2001). Further, regrowth on land affected by anthropogenic disturbances often takes different successional pathways from those typically found following natural disturbances, forming different biotas compared with the original habitats (Guariguata & Ostertag 2001; Lugo & Helmer 2004; Suding et al. 2004; Hobbs et al. 2006). Although we still have much to learn about the underpinning mechanisms that drive successional processes, human assistance may accelerate and lead to the successful recovery of the ecosystem as habitat for biodiversity (Parrotta et al. 1997; Yates & Hobbs 1997; Chazdon 2003).

The goals set for ecological restoration are broadly twofold: restoration of ecosystem function and diversity (Bradshaw 1984; Armstrong 1993; Lamb et al. 2005). Restoration of ecosystem function aims to recover ecosystem processes and services (e.g. water regulation, carbon sequestration, erosion control, pollination, nutrient cycling, food/ raw material production), whereas restoration of ecosystem diversity aims to recover the complex assemblages of (preferably native) species (Ehrenfeld 2000; Lamb et al. 2005). These goals are not mutually exclusive and achievement of both is necessary before an ecosystem can be considered fully restored (Bradshaw 1984, 1996). Other goals, such as timber production, aesthetic improvement and rescuing particular endangered species, are often incorporated into the broader goals of ecological restoration (Landres et al. 1988; Ehrenfeld 2000; Harrison et al. 2000; Swart et al. 2001).

Regardless of the goals of restoration, at the very least, successful development of vegetation (physical structure and plant species composition) is often used as a criterion for restoration success. However, the recovery of other ecosystem components, including soil characteristics and faunal composition, is also important in restored lands (Majer 1989b). Recovery of these ecosystem components is an interactive process and lack (or degradation) of one will impair the development of the others (Bradshaw 1984). In addition, both fauna and flora have their own aesthetic and intrinsic values that warrant their recovery, regardless of their roles in ecosystem functioning (Hobbs & Hopkins 1990; New 1993; Decaens et al. 2006). 2 Nevertheless, restoration projects have often focussed on the establishment of plants, with the underlying assumption that animals (and other native plants) will recolonise the rehabilitated sites naturally from surrounding areas, as succession of the vegetation proceeds (Majer 1990; Lamb et al. 2005), except in a limited number of cases where restoration management involves active manipulation of wildlife movements and demography (Scott et al. 2001). The bulk of earlier studies into the success of ecological restoration paid most attention to the development of vegetation or soil characteristics, with little attention to faunal colonisation in restored terrestrial habitats (Majer 1989b; Young 2000; Weiher 2007). However, even where physiognomy (e.g. canopy cover, plant structural diversity) and species composition of replanted vegetation may successfully emulate that of undisturbed habitat, this does not necessarily guarantee the return of native fauna, which may be limited by a site’s functional isolation from source habitats. Although an increasing number of recent restoration studies have focused on faunal communities (see Ruiz-Jaen & Aide 2005), there is still much to be discovered.

1.2 Restoration of cleared rainforest in Australia

In Australia, large tracts of vegetation have been cleared or heavily degraded as a result of extensive land use for agricultural production (Hobbs & Hopkins 1990; Australian State of the Environment Committee 2006). A recent report estimated that approximately 30% of Australia’s rainforests have been cleared since European settlement, mostly from lowland habitats (National Land and Water Resources Audit 2001). In north-east New South Wales, more than 99% of the subtropical rainforests that once covered 75 000 ha of the region have been cleared for grazing and other human use (Floyd 1990).

However, deforestation of tropical and subtropical rainforest in eastern Australia has all but ceased during the last few decades, and effort has begun on restoration of cleared landscapes (Erskine 2002; Catterall et al. 2004). Government funds, such as the Natural Heritage Trust schemes (Natural Heritage Trust 2000), have been used to implement a substantial number of revegetation projects aimed at restoring rainforest across the tropics and subtropics of Australia (Erskine 2002; Tucker et al. 2004; Australian State of the Environment Committee 2006; Catterall & Harrison 2006). Although representing only a fraction of what has been cleared in the past, around 2500 hectares had been revegetated with a variety of rainforest trees by 2002 (Catterall & Harrison 2006). Given a recent increase in the extent of abandoned land following intense agricultural 3 production, rainforest restoration in an old-field context will continue to be important in these regions. Restoration of cleared rainforest is, however, a relatively new endeavour with little known about successional patterns and mechanisms underpinning the re-establishment of biota in landscapes previously used for agriculture (Catterall et al. 2007).

1.3 Monitoring soil and litter arthropods in restored habitats

1.3.1 Reasons for monitoring soil and litter arthropods

To evaluate the performance of a restored habitat in terms of providing favourable conditions for native fauna, their colonisation patterns should be monitored through the implementation of regular sampling protocols (Block et al. 2001). In an ideal world, a monitoring program would cover the whole range of faunal groups that may colonise restored habitats in parallel with the developing vegetation (Andersen et al. 2004). However, restoration projects are invariably limited by financial resources (Clewell & Rieger 1997; Block et al. 2001), and restoration practitioners must therefore implement a cost-effective monitoring protocol that reliably reflects the state of restoration. Soil and litter arthropods are arguably some of the best candidates for use in monitoring protocols primarily because of (a) their significant diversity, (b) their strong association with ecological functioning, (c) their sensitivity to both biotic and abiotic factors and (d) their amenability to statistical analyses due to their high abundances. These are discussed below.

(a) Arthropod diversity. Invertebrates exhibit high abundance (biomass) and diversity, occupying almost the full range of microhabitats and niches (Wilson 1992; Kremen et al. 1993; Stork 1993). In particular, soil- and litter-dwelling invertebrates, most of which are arthropods (Decaens et al. 2006), are normally extremely abundant and diverse in terrestrial habitats (Ghilarov 1977; Wilson 1987, 1992; Andre et al. 1994). For example, in temperate woodland, soil with a well-developed litter and humus layer may contain more than a thousand invertebrate species, with overall abundances of several millions per square metre in just the top 5 cm (Swift et al. 1979). The high diversity of soil- and litter-invertebrates has been attributed not only to the heterogeneous characteristics of soil and litter (Anderson 1977) but also to the range of sizes of invertebrates that enable different species to occupy different niches within soil and litter (Ghilarov 1977; Kaspari

4 & Weiser 1999; Sarty et al. 2006). Despite their diversity, soil and litter invertebrates have not received much attention in the ecological restoration literature until recently (Andre et al. 2002; Decaens et al. 2006). Giller (1996) suggested that this lack of attention was perhaps the result of the cryptic nature of the soil-litter system and an absence of aesthetically attractive species.

(b) Association with ecological functioning. Soil and litter arthropods are known to play an essential role in many ecological processes (e.g. predation, parasitism, herbivory, soil formation, decomposition), thus their presence or absence may be a useful indicator of the development of ecological processes in restored habitats (Majer 1989a; Curry & Good 1992; Lavelle et al. 2006). In particular, their contribution to decomposition of organic matter, an essential part of nutrient cycling, may be important (Reichle 1977; Swift et al. 1979; Giller 1996; Davies et al. 1999; Hansen 1999; Heneghan et al. 1999). Soil and litter arthropods contribute to this decomposition both directly and indirectly (Giller 1996). Direct decomposition involves ingestion and chemical breakdown of litter components by detritivorous invertebrates (Giller 1996). Indirect decomposition includes: physical breakdown (comminution) of litter and production of faecal pellets by arthropods that enhance leaching of nutrients and decomposition by soil microbes (Seastedt & Crossley 1980; Dangerfield 1996; Prescott 2005); and the enhancement of microbe activities through regulation of the soil microbial community by grazing, disturbance and dispersal of arthropods (Standen 1978; Seastedt & Crossley 1984; Moore et al. 1988). Many arthropod groups including mites, millipedes, isopods, and amphipods are involved in decomposition processes both directly and indirectly (Hopkin & Read 1992; Lussenhop 1992; Maraun et al. 1998; O'Hanlon & Bolger 1999; Walter & Proctor 1999; Smit & van Aarde 2001). Other arthropod groups, notably ants, are involved in a range of other key ecological processes, such as seed dispersal and soil formation, and are functionally important at various trophic levels (Buckley 1982; Matthews & Kitching 1984; Holldobler & Wilson 1990; Folgarait 1998; Boulton & Amberman 2006).

The extent of arthropod diversity required to maintain such ecological processes is poorly understood (Kremen 2005; Wardle 2006). This is because the contribution of a given set of organisms to ecosystem processes is governed by complex factors including individual species’ trait differences, interactions among species, and temporal and spatial variations (Loreau et al. 2001). Given our limited understanding of these factors, sustainable ecological processes are arguably best supported by maximising the diversity of 5 arthropod species and functional groups that directly relate to ecosystem processes (Walker 1992; Bengtsson 1998; Tilman et al. 2006).

(c) Sensitivity to biotic and abiotic factors. Changes in biotic and abiotic attributes are known to influence the distribution, diversity and abundance of arthropods (Collins & Thomas 1991; Didham et al. 1996; Wheater et al. 2000; Willett 2001; Pik et al. 2002; Woinarski et al. 2002). Because many soil and litter arthropods are habitat specialists with short generation times and sedentary natures, they may reflect local habitat conditions better than other long lived, highly mobile organisms, such as vertebrates and flying invertebrates (van Straalen 1997, 1998; Hilty & Merenlender 2000). It has been suggested that the organisation of arthropod assemblages is chiefly driven by localised habitat conditions (e.g. microclimate, substrate conditions), rather than landscape-level habitat characteristics (e.g. configuration of the surrounding matrix) (Niemela et al. 1996; Dauber et al. 2005; Grimbacher et al. 2006). The composition of soil and litter arthropod assemblages are affected by many attributes, including floristic diversity and structure (Decaens et al. 1998; Armbrecht & Perfecto 2003), temperature (Torres 1984; Shure & Phillips 1991; King et al. 1998; Botes et al. 2006), soil and litter humidity (Levings & Windsor 1984; Shure & Phillips 1991; Blake et al. 2003; Lassau & Hochuli 2004; Chikoski et al. 2006), soil chemistry and structure (Boulton et al. 2005) and quantity and composition of litter (Kinnear 1991; Sousa & da Gama 1994; Decaens et al. 1998; Hansen 1999; Armbrecht et al. 2004).

(d) Amenability to statistical analyses. A number of authors have attested that the ubiquitous occurrence and large abundance and species richness of arthropods make them amenable to statistical analyses, unlike other organisms such as large-bodied vertebrates that are often only sparsely present (Cranston & Trueman 1997; Basset et al. 1998). Also, their great abundances make sampling easier as only relatively small sampling intensity and duration are required to yield adequate numbers of individuals for statistical analysis. Moreover, fewer social and ecological considerations constrain the collection of terrestrial invertebrates (Kremen et al. 1993). However, despite the ubiquitous occurrence of arthropods, individual species may be rare or patchily distributed, even within a habitat with homogeneous environmental conditions (Wilson 1958; Novotny & Basset 2000; Soares & Schoereder 2001), requiring a carefully designed arthropod sampling protocol.

6 It should be noted that even if soil and litter arthropods prove useful for monitoring ecological restoration, this does not devalue the use of other faunal groups. Some restoration studies have suggested that certain groups of animals may reflect the responses of other taxa in restored habitats, that is, one taxon can be used as a surrogate for others (Majer 1983, 1996). However, the concept of surrogacy has been questioned in many fields of ecology, including conservation management (Kareiva 1993; Lawton et al. 1998; Moritz et al. 2001; Schulze et al. 2004) and rapid biodiversity assessment (Cranston & Trueman 1997; Kotze & Samways 1999; Alonso 2000; Ward & Lariviere 2004). Recent studies into the restoration of former mine sites in sclerophyllous forest in Australia (Nichols & Nichols 2003) and coastal dune forest in South Africa (Wassenaar et al. 2005) support this view. These authors found that, although assemblage compositions of different faunal groups in restored habitats were predicted to eventually converge with those of undisturbed habitat, the recolonisation pattern of one taxon (e.g. rate of recolonisation) does not necessarily reflect that of others (see also Majer & Nichols 1998; Bisevac & Majer 2002).

1.3.2 Approaches to cost-effective monitoring with soil and litter arthropods

In spite of these advantages associated with using arthropods as indicators of environmental change, there are also two significant challenges to doing so cost-effectively. First, samples of soil and litter biota require time consuming post-field processing, and second, there is a lack of biological, ecological and taxonomic knowledge for most arthropod groups (Lawton et al. 1998; Landsberg et al. 1999). Lack of taxonomic knowledge may be one of the major constraints in studying the ecology of arthropods. Given the dwindling funds for traditional taxonomic studies and an associated decline in the number of specialised taxonomists, it is becoming less likely that taxonomic studies will be completed, if ever, for large numbers of undescribed taxa (Gaston & May 1992; New 1996; Wilson 2000; Godfray 2002; Samyn & Massin 2002; Wheeler et al. 2004; Kim & Byrne 2006). In order to deal with this taxonomic impediment (sensu Kitching 1993b) as well as to reduce the costs associated with sorting, three approaches have been used in ecological studies: (i) the use of higher taxonomic resolutions, (ii) the use of morphospecies (also known as ‘recognisable taxonomic units’ or ‘operational taxonomic units’), and (iii) the use of focal groups, as ‘ecological indicators’ (sensu McGeoch 1998) to provide information about the state of complex ecosystems.

7 Use of higher taxonomic resolutions Identification of organisms at supra-species taxonomic resolution (e.g. genus, family, order) requires less taxonomic knowledge, freeing up time and resources that would otherwise have been used for species identification (Williams & Gaston 1994; Balmford et al. 1996). In addition, higher-taxonomic ranks often broadly reflect different functional guilds, which enhances understanding of community organization from an ecological point of view (Hawkins & Macmahon 1989; Simberloff & Dayan 1991; Andersen 1995a; Churchill 1997; Uetz et al. 1999). The use of a higher-taxon approach has been tested extensively using a diverse array of organisms including birds, bats, arthropods, fish, amphibians and plants across various aquatic and terrestrial habitats (Gaston 2000; Hughes et al. 2000; Gladstone & Alexander 2005; Villasenor et al. 2005). Provided that an appropriate taxonomic rank is chosen, many authors have advocated the use of a higher-taxon approach to identify conservation priorities or to detect changes in habitat conditions (Williams & Gaston 1994; Gaston 2000; Pik et al. 2002; Baldi 2003; Caruso & Migliorini 2006). However, others have raised concerns that the reliability of a higher-taxon approach may vary among different groups of organisms and may mask important information that would have been obtained if organisms were sorted at finer taxonomic resolutions (e.g. Andersen 1995b; Hirst 2006; Quijon & Snelgrove 2006).

Unless parallel studies using finer levels of taxonomic resolution are carried out, results obtained using a higher taxonomic approach should be treated with caution. For example, Pik et al. (2002) evaluated the responses of soil and litter arthropods at different taxonomic ranks across sites subjected to different restoration treatments. They found that, regardless of the taxonomic resolution used, arthropod assemblages clearly responded to obvious habitat changes: assemblages from undisturbed habitat significantly differed from those of all restored habitats. Within the restored sites, however, only arthropod assemblages sorted at least to genus showed clear differences in their assemblage structures, separating sites under different restoration treatments, whereas arthropods sorted to Order failed to do so. In this example, the use of ordinal data alone may have drawn the misleading conclusion that there were no differential effects of restoration treatments on the development of soil and litter arthropods.

8 Use of morphospecies Unlike other approaches, sorting arthropods to morphospecies requires little taxonomic knowledge. Morphospecies are assigned by non-specialists or para-taxonomists using morphological criteria with little or no reference to taxonomic classifications (Beattie & Oliver 1994). This approach may save substantial costs as requisite collaboration with specialist taxonomists is minimal or even nonexistent, and has therefore been used extensively (e.g. Hughes et al. 2000; Basset et al. 2004; Mathieu et al. 2004; Kattan et al. 2006). A number of studies have evaluated the efficacy of a morphospecies approach to monitor the response of arthropod assemblage composition across different habitat characteristics (Oliver & Beattie 1993, 1996a; Oliver & Beattie 1996b; Pik et al. 1999; Pik et al. 2002; Ward & Stanley 2004). Oliver and Beattie (1996a), for example, tested the use of morphospecies as surrogates for species using assemblages of ants, beetles and spiders across different forest types. They found that both morphospecies and species of all three arthropod groups separated major forest types in multivariate ordinations of assemblage compositions. Despite this, the use of morphospecies, especially as a substitute for species-level identification by specialist taxonomists, has been severely criticised for a number of reasons. First, depending on the particular group of organisms studied, as well as the skill of an individual sorter, discrepancies in the number of entities between morphospecies and species-level taxonomic sorting can be as high as 100% (Campbell 1995; Krell 2004). Beattie and Oliver (1994) argued that this shortfall can be amended by trial tests comparing results obtained by morphospecies- and species-level sorting. However, this evaluation process itself takes time and money, conflicting with the underlying rationale of this approach (Brower 1995). Second, loss of species-specific information usually prevents cross-referencing between studies conducted at different times or locations (Ward & Stanley 2004). In addition, a morphospecies approach prevents the identification of patterns in attributes such as species’ functional significance, rarity and endemicity, impeding prioritisation of locations where conservation and management attentions are most needed (Goldstein 1997; Trueman & Cranston 1997; Goldstein 1999).

Use of focal groups As an alternative to monitoring the whole (or a large subset of) arthropod assemblages, certain arthropod taxa may be selected as focal groups with the hope that they reliably indicate changes in habitat conditions, and/or represent the composition of other faunal

9 groups (Andersen 1990; Giller 1996; Churchill 1997; Rainio & Niemela 2003; Andersen & Majer 2004; Cardoso et al. 2004; Pearce & Venier 2006). Although this approach requires extensive knowledge of the taxonomy of the focal arthropod groups, sorting costs may be reduced as detailed examination of specimens is restricted to those groups.

With regard to soil and litter arthropods, a number of taxa, namely ants (Majer 1983; Agosti et al. 2000; Andersen & Majer 2004), beetles (McGeoch et al. 2002; Rainio & Niemela 2003; Pearce & Venier 2006), spiders (Churchill 1997; Willett 2001; Pearce & Venier 2006), certain groups of mites (Franchini & Rockett 1996; Ruf 1998; Beaulieu & Weeks 2007) and collembolans (Greenslade 2007), have been suggested as possible ecological indicators of habitat changes. These taxa are chosen not only because they meet the characteristics required for use in monitoring (i.e. their diversity, association with ecological functioning, sensitivity to biotic and abiotic factors, amenability to statistical analyses), but also because their natural history and taxonomy are relatively well known compared with other arthropod groups (Pearson 1994). These focal taxa, however, have generally been used only over limited geographic regions and their effectiveness as indicators has not been demonstrated elsewhere. For example, a large body of studies has advocated the use of carabid beetles (Coleoptera: Carabidae) as reliable ecological indicators, but these studies are chiefly from North America and Europe (see Rainio & Niemela 2003). In an Australian context, carabid beetles may not be so useful, as their abundance and diversity in a given area are often low, and little is known about the ecology and taxonomy of the Australian species (New 1998). When choosing a focal taxon, therefore, it may be sensible to choose groups whose efficacy as ecological indicators has been demonstrated extensively in the region of interest.

In Australia, ground-dwelling ants have been the most extensively-used arthropod indicators of habitat changes (see Hoffmann & Andersen 2003). Many studies have advocated the use of ants because of their relatively well-known ecology, taxonomy, ubiquitous occurrence, and most importantly, their sensitivity to changes in biotic and abiotic factors (Greenslade & Greenslade 1984; Andersen 1990; Agosti et al. 2000; Andersen et al. 2004; Andersen & Majer 2004). In Australia and elsewhere, ants have been monitored in a wide range of habitats, and their usefulness as ecological indicators been consistently demonstrated (see Andersen & Majer 2004). In particular, responses of ant assemblages to natural and human-induced disturbances are well documented and understood (Hoffmann & Andersen 2003). The types of disturbance studied include 10 hurricane damage (Morrison 2002), livestock grazing (Andersen & McKaige 1987; Andersen 1991a; Bestelmeyer & Wiens 1996; Hoffmann 2000), agricultural activities (Keals & Majer 1991; Lobry de Bruyn 1993; Roth et al. 1994; Peck et al. 1998; Perfecto & Vandermeer 2002; Gomez et al. 2003; Armbrecht et al. 2005), logging (Neumann 1991; Punttila et al. 1991; Basu 1997; Vanderwoude et al. 2000a; Vasconcelos et al. 2000; Kalif et al. 2001; Azevedo-Ramos et al. 2006; Palladini et al. 2007), natural and prescribed fire (Neumann 1991; Vanderwoude et al. 1997; Andrew et al. 2000; York 2000; Stephens & Wagner 2006), pollution (Koponen & Niemela 1995; Hoffmann et al. 2000; Andersen et al. 2002), urbanization (Burbidge et al. 1992) and mining (see Appendix 1).

For more effective and ecologically meaningful monitoring of arthropod communities, individual taxa or species are often assigned to functional groups based on ecological rather than taxonomic criteria (Terborg & Robinson 1986; Humphries et al. 1995). A functional group approach may allow comparison of patterns of community structure found in different habitats independent of species identity, allowing more general interpretations than would be possible using a taxon-level approach (Terborg & Robinson 1986; Didham et al. 1996; Andersen 1997b). For studies of terrestrial arthropods, feeding guilds (e.g. predator, herbivore, omnivore, fungivore) have been used in conjunction with higher-taxon or species level analyses (Moran & Southwood 1982; Hawkins & Macmahon 1989). With regard to ants, however, only a limited group of taxa have specialised food resources, and most are generalised omnivores (Brown 2000), making the use of a feeding guild scheme impractical (see Andersen 1991c).

Australian studies of ant assemblage organization have generated a different functional group scheme that categorises ants on the basis of interspecific interactions and ecological and behavioural traits (Greenslade 1978; Greenslade & Halliday 1983; Andersen 1990; Andersen 1995a; Andersen & Majer 2004). This classification scheme was first proposed by Greenslade (1978) for ant assemblages in arid environments, and was later improved by Andersen (1990; 1995a; 1997b). Ant functional groups consist of three major, highly interactive groups (viz. Dominant Dolichoderinae, Generalised Myrmicinae, Opportunist) and other presumably less interactive ones that are categorised on the basis of their subordinate or cryptic behaviour, specialised food resources or certain climatic associations (Table 1.1). Relative abundances of the three major functional groups are predicted to vary according to the presence and severity of 11 environmental stress (factors limiting productivity) and disturbance (factors removing biomass) (Andersen 1995a). Increased severity of stress is predicted to favour cosmopolitan ant genera belonging to the Generalised Myrmicinae, whereas disturbance is predicted to cause proliferation of unspecialized Opportunists. In the absence of stress and disturbance, competitive interaction is predicted to become a primary factor, favouring highly active and aggressive species of Dominant Dolichoderinae that regulate the organization of ant assemblages. Approaches using ant functional groups have been used in a variety of habitats and climatic regions in Australia as well as other regions of the world, including South Africa, North and South America and Europe (Bestelmeyer & Wiens 1996; Andersen 1997b; Gomez et al. 2003; van Hamburg et al. 2004; Ottonetti et al. 2006; Stephens & Wagner 2006, see also Table 1.1).

1.3.3 Studies of soil and litter arthropods in actively-restored lands

Through the second half of the last century into the new millennium, an increasing number of studies have investigated the successional processes of terrestrial restoration using a wide range of organisms (Butcher et al. 1989; Young et al. 2005). Although small relative to the entire body of restoration literature, over 50 studies have investigated soil and litter arthropods colonising actively-restored terrestrial habitats (see Appendix 1). In this section I summarise the nature and findings of 53 studies, with particular emphasis on the type of restoration, study design and factors ultimately considered important for the development of assemblages of soil and litter arthropods. This review comprises journal papers or reports written in English that have presented original data obtained from actual terrestrial restoration involving active human intervention, such as soil amelioration, seeding and/or replanting. Studies investigating natural regrowth without human intervention have not been included.

12 Table 1.1 Summary of ant functional groups with some of the major genera found in Australia, North America or Europe (after Andersen 1997b; Gomez et al. 2003). See also Brown (2000) for the ant genera of the world with their associated functional groups. Functional group and major genera Characteristics Dominant Dolichoderinae Highly active and aggressive, exerting a strong competitive influence on Iridomyrmex, Anonychomyrma (Australia) other groups of ants. Tapinoma (Europe) Forelius, Liometopum (Nth America) Generalised Myrmicinae Cosmopolitan genera occurring in most habitats. Rapid recruitment to, and Pheidole, Crematogaster, Monomorium (part) (all three regions) successful defence of, clumped food resources. Opportunists Unspecialized species characteristic of disturbed sites, or other habitats Rhytidoponera, Paratrechina, Ochetellus (Australia) supporting low ant diversity. Myrmica, Formica (part) (Nth America, Europe) Cryptic species Forage predominantly within soil and litter, having relatively little Solenopsis, Hypoponera (all three regions) interaction with epigaeic ants. Subordinate Camponotini Co-occurring with, and behaviourally submissive to, dominant Camponotus (all three regions) dolichoderines. Specialist predators Little interaction with other ants due to specialized diet, large body size Leptogenys, Pachycondyla, Platythyrea (Australia, Nth America) and small colony size. Hot climate specialists Arid-adapted species with morphological, physiological or behavioural Melophorus, Meranoplus, Monomorium (part) (Australia) specializations which reduce their interaction with dominant Messor, Myrmecocystus, Pogonomyrmex (Nth America) dolichoderines. Messor, Cataglyphis (Europe) Cold climate specialists Distribution centred on the cool-temperate zone. Occur in habitats where Monomorium (part), Notoncus, Prolasius (Australia) dominant dolichoderines are generally not abundant. Lasius , Leptothorax (Nth America, Europe) Tropical climate specialists Distribution centred on the humid tropics. Occur in habitats where Oecophylla, Leptomyrmex, Lordomyrma (Australia) dominant dolichoderines are generally not abundant.

13 Neivamyrmex, Pseudomyrmex (Nth America)

Type of restoration Most studies of active restoration which incorporated soil and litter arthropods focused on revegetated former minesites (38 studies, Table 1.2). Unlike restoration of other habitat disturbances, revegetation of mined sites and monitoring of rehabilitation performance are mandatory under mining legislation in countries such as Australia (e.g. NSW Dept. of Primary Industries’ Mining Act 1992), the USA (U.S. Federal Government’s Surface Mining Control and Reclamation Act of 1977) and South Africa (Mineral and Petroleum Resources Development Act). Only nine studies involve the active restoration of old fields (i.e. abandoned agricultural land). Even fewer studies (six) have been conducted in other types of rehabilitation, which include restoration of cleared vegetation (for timber or as a construction site), landfill and disused roads (Appendix 1). Unlike minesites that have a long history of revegetation, other types of restoration (including old-field restoration) are relatively new endeavours with most studies conducted within the last 10 years (1998-2007) (Appendix 1).

Table 1.2 The number of studies of actively-restored vegetation which incorporated soil and litter arthropods, carried out in different regions and habitat types. Studies are subdivided into minesite restoration, old field restoration and other types of restoration. Note that the total number of studies from habitat types does not match the actual number of studies summarized as one study often straddled two or more habitat types. Minesite Old field Others Total Total number of studies 38 9 6 53 Region Australasia 17 6 3 26 Europe 9 1 0 10 Nth America 5 1 2 8 Africa 5 0 1 6 Sth and Central America 2 1 0 3 Asia 0 0 0 0 Total 38 9 6 53 Habitat type Sclerophyllous forest 13 1 1 15 Temperate forest† 10 2 0 12 Shrubland 5 2 1 8 Rainforest 2 3 1 6 Grassland 4 0 2 6 Dune forest 5 0 0 5 Heathland 3 1 0 4 Riparian forest 0 0 2 2 Montane forest 0 1 0 1 Marsh 0 0 1 1 Total 42 10 8 60 † Temperate forest includes both deciduous and coniferous forests. 14 Study regions and habitat types Most studies of active restoration involving soil and litter arthropods have been conducted in a limited number of regions, namely Australasia (Australia, New Zealand), Europe and North America (Table 1.2). While the majority of all restoration studies come from North America (Butcher et al. 1989; Young 2000; Ruiz-Jaen & Aide 2005), nearly half of the studies involving soil and litter arthropods are from Australasia. Although several studies have been carried out recently in Africa, all but one were located in the same area, KwaZulu Natal, South Africa. Few studies have been carried out in South and Central America, possibly because natural regrowth of secondary forest occurs much more commonly, and human intervention is less commonly involved (Brown & Lugo 1990; Brown & Lugo 1994; Lugo & Helmer 2004). No studies of this type have been carried out in Asia, despite a recent net gain in the extent of forest cover as a result of large-scale reforestation reported in China (FAO 2006).

Most of these studies have been conducted in sclerophyllous forests in Australia and temperate deciduous/coniferous forests primarily in Europe. In both regions, mining and subsequent restoration is commonplace. Other habitat types have been studied to a much lesser extent (Table 1.2).

Target taxa and taxonomic resolutions Among arthropods, ants, beetles, spiders, mites and springtails have been commonly investigated within studies of active restoration involving soil and litter arthropods (Table 1.3, see also Appendix 1). In particular, ants have been incorporated most often. This is partly because ants are one of the most abundant and diverse groups of soil and litter invertebrates in arid areas of Australia, where many minesites have been revegetated and studied (e.g. Majer 1983; Majer et al. 1984; Majer 1985; Andersen 1993; Majer & Nichols 1998; Bisevac & Majer 1999). In cooler and wetter regions, less attention has been paid to ants, and other groups of arthropods, particularly beetles, have been investigated more commonly.

In addition to numerous species-level studies, other higher taxa – genus, family and Order – have also been used (Table 1.3), particularly when all or a large subset of arthropods have been sampled and analysed. Species-level analyses have been employed commonly when ants or beetles were chosen as focal taxa. A functional group approach (feeding guilds, ant functional groups) has also been employed by a number of studies. About half

15 (10 of 25 studies) of the studies that investigated ants at the species-level of taxonomic resolution also employed ant functional groups.

Table 1.3 The number of studies of actively-restored vegetation incorporating groups of soil and litter arthropods at different taxonomic and functional ranks. Note that one study often investigated two or more groups of arthropods with various taxonomic and functional ranks. Taxonomic rank Functional group Morpho- Other higher Ant functional Feeding

Species species taxon groups guild Total Ants 25 4 12 10 4 55 Beetles 11 3 15 n/a 5 34 Spiders 6 2 15 n/a 5 28 Springtails 8 1 13 n/a 3 25 Mites 6 2 15 n/a 2 25 Others† 12 1 15 n/a 3 31 Total 68 13 85 10 22 †Other groups of arthropods include centipedes, millipedes, isopods, orthopterans and dipteran larvae.

Sampling methodology Pitfall traps are relatively cheap and easy to install, hence are used by a large number of scientific studies that aim to sample the activities of epigaeic arthropods (Luff 1975; Holland & Smith 1999; Pekar 2002). Indeed, pitfall trapping is the most commonly used sampling method in studies of soil and litter arthropods within actively-restored vegetation (39 of 53 studies, Appendix 1). Pitfall traps, however, generally collect arthropods that actively forage on the ground. More sedentary arthropods that forage mostly in soil and litter tend to be under-sampled by this trapping method (Fisher 1999). In addition, preserving agents, such as ethanol and ethylene glycol, used in pitfall traps tend to attract or repel certain groups of arthropods, causing a bias in captures of some taxa (Greenslade & Greenslade 1971; Adis 1979; Pekar 2002). Despite these shortfalls, pitfall traps do provide comparative information on species richness and relative abundances of actively moving epigaeic arthropods as long as sampling protocols are standardised across sites (Andersen 1990). Many restoration studies have employed pitfall trapping to target ants and beetles that actively forage on the ground (Appendix 1).

As an alternative to the use of a single trapping method, sampling is often complemented by one or more additional methods. A combination of sampling methods provide more comprehensive information on the distributions of arthropods, particularly when habitats

16 are complex and a single collection method cannot adequately sample arthropods from various types of microhabitats such as forest litter, decomposing logs and soil substrate (Delabie et al. 2000). In the restoration studies summarised here, 20 of 53 studies employed two or more sampling methods. In all but one of the 20 studies, pitfall trapping was complemented by other sampling methods (e.g. hand collection, visual survey, litter extraction, see Appendix 1).

Reference habitats To evaluate the levels of restoration success, the composition of developing biota is often compared with surveys of reference sites which provide milestones against which changes can be evaluated (Parker & Pickett 1997). Reference information can be categorised into two broadly defined sets in the context of successional pathways – one representing the state that restoration projects aim ultimately to achieve (i.e. undisturbed systems) and the other representing the state that existed before restoration commenced (i.e. disturbed systems) (Ruiz-Jaen & Aide 2005). Reference information may be sourced from historical records of the same location before and after anthropogenic disturbances took place. However, comprehensive sets of historical information are rarely available for most studies. Furthermore, historical data may not necessarily represent achievable goals in the future given that most ecosystems are dynamic, changing their characteristics as a result of large scale anthropogenic disturbances (e.g. global warming) or other environmental changes (Hobbs & Harris 2001; Harris et al. 2006). Restoration studies, therefore, generally utilise contemporary data collected from nearby undisturbed and disturbed sites considered to be good analogues of the restoration goal (White & Walker 1997). Indeed, contemporary data were utilised by all of the summarised studies that collected reference information form their ‘control’ sites.

In 46 out of the 53 studies summarised here, at least one undisturbed site was chosen as reference habitat (Table 1.4). This, however, was not the same for disturbed sites: only 11 studies included one or more sites that represent the pre-restoration disturbed state of the habitat. The absence of disturbed reference sites is perhaps justifiable for the study of minesite restoration as disturbed (mined) sites are presumably devoid of most organisms. However, in many other types of terrestrial restoration, including old field restoration, a certain set of biota may persist or proliferate in the disturbed sites such as pasture and cropland (Beare et al. 1997; Green & Catterall 1998). Therefore, undertaking

17 measurements of the biota in such disturbed reference habitats provides an important basis for establishing the changes following restoration.

It has been suggested that, where possible, these reference habitats be spatially and temporally replicated to capture natural variations of biota through space and time (Parker & Pickett 1997; White & Walker 1997). Spatial replication across the study area is needed because many groups of organisms, particularly arthropods, are patchily distributed even if the area appears to be environmentally homogenous (e.g. Franks & Bossert 1983; Grear & Schmitz 2005). However, most of the summarised studies included one or, at most, two spatially replicated reference sites (Table 1.4), unless there were clearly different types of habitat present within a study area (e.g. different forest types).

Table 1.4 The number of studies of actively-restored vegetation involving soil and litter arthropods that did or did not include reference sites, and the number of spatial replications employed. Reference sites were subdivided into either ‘undisturbed’ (i.e. intact habitats or old secondary-growth forests, generally representing the ‘goals’ of restoration projects), or ‘disturbed’ (habitat representing the status before restoration commenced). Undisturbed Disturbed No of Number of reference sites used 0 1 2-3 >3 0 1 2-3 >3 studies Minesite 7 12 16 3 33 5 0 0 38 Old field 0 4 2 3 4 3 0 2 9 Others 0 3 2 1 5 1 0 0 6 Total 7 19 20 7 42 9 0 2 53

Understanding temporal variation not only helps estimate ecological variation of reference sites through time but also helps evaluate whether a restored site achieved a self-sustaining state (White & Walker 1997). Such temporal replication of reference sites may encompass both short- (i.e. seasonal variation) and long-term (i.e. interannual variation) dynamics of biota (White & Walker 1997). While short-term variation can be readily measured, the measurement of long-term variation requires longitudinal studies that repeatedly collect samples from the same site over a long period of time. Most of the summarised studies (41 of 53), however, employed a chronosequence approach that took ‘snap-shots’ of restored sites, each of which generally represented different ages of restoration (Appendix 1). There were 12 longitudinal studies, nine of which included at least one reference site, with temporal replication spanning more than one year (Majer

18 1981; Van Dijk 1986; Williams 1993; Simmonds et al. 1994; Williams 1997; Majer & Nichols 1998; Bisevac & Majer 1999; Nichols & Nichols 2003; Schnell et al. 2003, see Appendix 1). Longitudinal studies may reveal important temporal variations that might otherwise be missed by a chronosequence approach. Nevertheless, this approach requires long-term commitment to a study, and spatial replication is often compromised due to the substantially increased cost needed to sample one site repeatedly (Majer & Nichols 1998).

Factors affecting patterns of arthropod colonization In order to guide and accelerate the progress of restoration towards the state of the reference habitat, it is necessary to understand the underlying factors that facilitate or impede development of organisms. Using empirical as well as anecdotal evidence, previous studies have investigated a large number of biotic and abiotic factors that may directly or indirectly influence the development of arthropods. Table 1.5 summarises the types of factors investigated by studies of the occurrence of soil and litter arthropods in actively-restored habitats.

The effect of time since restoration commenced (restoration age) has been by far the commonest factor investigated, and most studies reported that species richness and abundances of arthropods increased with the age of restoration (Table 1.5). In general, older restoration sites have more favourable habitat conditions, with more developed vegetation structures and greater plant species richness, than do younger sites. Further, the chances of successful colonisation are increased by virtue of their greater age.

Many other factors, particularly those associated with restoration management, floristic attributes and litter and soil characteristics, have been relatively well studied. A number of factors (viz. restoration techniques, plant taxonomic diversity, canopy cover, ground covered by vegetation, tree height, litter depth/amount, ground covered by litter) have been investigated by more than 10 studies, with more than half reporting significant or potential effects on colonisation patterns of soil and litter arthropods (Table 1.5). In general, floristic attributes (both structural and diversity characteristics) appeared to be associated more often with variations in assemblage composition of soil and litter arthropods than either litter or soil characteristics.

19 While patterns of arthropod recolonisation will ultimately be affected by uncontrollable factors (e.g. restoration age, climate, inter-species interactions, surrounding matrix and natural disturbances), many of the factors listed in Table 1.5 are potentially under the control of those conducting restoration. Increased plant taxonomic diversity, for example, was found important for the development of arthropods by most studies (Table 1.5). Although plant taxonomic diversity may increase simply with time through natural colonisation, this can be augmented by planting a diverse array of plant species. Similarly, full establishment of canopy cover may be achieved within a short period of time if canopy trees are closely planted.

To date, studies of actively-restored vegetation which incorporated soil and litter arthropods have all employed post-hoc empirical observation of existing restoration sites. This approach presents challenges in the design of any replicated study, as restoration projects are generally carried out on a site-specific basis with no spatial replication, undermining statistical inferences from the survey data (Block et al. 2001). Furthermore, many characteristics, such as time since establishment, patch size, plant species composition and proximity to other remnant rainforest, vary greatly and co-vary in an uncontrolled fashion from one site to another (Michener 1997; Catterall et al. 2004), which makes investigation of individual factors difficult. Multicollinearity of investigated factors is a common problem in many field of ecological study (Graham 2003), and restoration studies are no exception (see also Majer et al. 1984; van Hamburg et al. 2004).

20

Table 1.5 The number of studies investigating the effect of various factors on the occurrence of soil and litter arthropods in actively-restored terrestrial habitats. Included factors were investigated by at least two restoration studies. When a given factor was found to influence colonisation patterns of arthropods, it was tallied into either: ‘significant’ when that factor was tested by means of statistical analyses and/or graphical representations; or ‘potential’ when only anecdotal evidence was provided. Litter

Floristic attributes characteristics Soil characteristics Others

Restoration age Restoration techniques Plant taxonomic diversity Canopy cover (%) Ground covered by vegetation (%) diversity Plant structural Tree height Tree density Litter depth/amount Ground covered by litter (%) Coarse woody debris composition Litter Soil chemistry (C, N, P, pH) Soil penetrability Soil structure content Soil moisture rocks Presence of interactions Inter-species Ground temperature Distance from undisturbed habitats temp) (rainfall, Climate Pollution significant 18 8 11 6 5 4 2 1 6 6 2 1 3 1 1 1 0 4 2 1 0 1 potential 19 12 6 8 7 5 4 1 8 4 4 1 4 3 3 3 0 2 2 3 4 1 not significant 5 1 3 0 10 0 5 0 7 8 6 0 3 6 1 5 2 0 5 1 0 0 Total 42 21 20 14 22 9 11 2 21 18 12 2 10 10 5 9 2 6 9 5 4 2 21

Differences between minesites and other types of restoration As shown in Table 1.2, restoration studies of soil and litter arthropods are primarily from revegetated minesites. Although these studies have significant implications for other types of restoration, there are key differences in minesite restoration that should be recognised.

First, the nature and severity of disturbance in mined sites is different from that following other forms of land use. Mining activities generally remove the entire assemblage of inhabiting biota by removing topsoil following clearing of vegetation (Majer 1989b). Mining also causes further damage by excavation and mixing of overburden. Although post-mining treatment may include soil amelioration, such as liming, replacement with fresh or stored topsoil and deep ploughing (Bradshaw 1983; Majer et al. 1984), substrate conditions remain strikingly different from those in undisturbed habitat or in cleared but unmined land (Davis 1963; van Aarde et al. 1998; Dunger et al. 2001). The severity of the disturbance in mine sites was demonstrated by Jackson and Fox (1996) who found slower recovery of ant assemblage composition in mined sites compared with that in cleared, but unmined sites. In other types of terrestrial restoration, the nature of ecological damage is generally different from that of mined sites (Hobbs 1993). For example, old field restoration is generally carried out in former pastoral land. Although livestock grazing may have caused extensive ecological damage (Smith 1940; Wilson 1990; Fleischner 1994; Read 1999; Clapperton et al. 2002), the areas targeted for restoration are generally vegetated and contain a range of arthropod species (though assemblage composition may be very different from comparable undisturbed habitats).

Second, while minesites are generally surrounded by a large amount of relatively intact forest that act as a source of recolonising fauna, other types of restoration are often surrounded by a large extent of disturbed matrix, with limited presence of nearby intact habitats (Hobbs 1993). In nine of the 10 summarised studies of minesite restoration that described surrounding landscapes, intact forest was present near the study area (Neumann 1973; Fox & Fox 1982; Majer & de Kock 1992; Andersen 1993; van Aarde et al. 1996a; St. John et al. 2002; Davis et al. 2003; Nichols & Nichols 2003; Ottonetti et al. 2006). In contrast, the presence of nearby intact habitat seems to be highly variable in other types of restoration. For example, in six of the nine studies of old field restoration, the restored sites were mostly surrounded by a matrix of pasture, urban land or other disturbed

22 habitats (King et al. 1998; Watts & Gibbs 2002; Longcore 2003; Nakamura et al. 2003; Schnell et al. 2003; Grimbacher et al. 2007), whereas three study sites were surrounded by, or adjacent to, intact habitats (Van Dijk 1986; Jansen 1997; Kattan et al. 2006). Unlike highly mobile organisms (e.g. flying insects, Grimbacher & Catterall 2007), colonisation patterns of less-mobile organisms, such as soil and litter arthropods, may be highly affected by isolation of restored sites.

1.4 Aims and thesis structure

The broad objective of this thesis is to investigate the effects of selected factors on the development of soil and litter arthropod assemblages in actively-restored habitat patches. The intention of the project was to provide knowledge that would help identify ways in which restoration practitioners could adjust the design of their revegetation projects so as to facilitate colonisation by arthropods characteristic of undisturbed natural habitats. The entire project was carried out in the context of restoration of land following rainforest clearance and associated land use for pasture. Factors were selected for study on the basis of their potential influence on colonisation and establishment patterns of arthropods but were limited to those that could potentially be controlled through restoration management.

The studied factors are as follows:

• effect of isolation (distances between restored habitat patches and a source of potential colonisers);

• efficacy of inoculation (re-introduction of soil and litter arthropods from within remnant rainforest) in isolated habitat patches where natural colonisation by arthropods is unlikely;

• effects of quality and quantity of mulch used during the initial stages of rainforest restoration;

• effects of shading and litter (mulch) depth; and,

• impacts of glyphosate herbicide on soil and litter arthropods potentially colonising restored habitat patches.

23 In order to test explicitly the effects of individual factors, an experimental approach was employed in this project. The first four factors were addressed by means of a manipulative field experiment in which small-scale habitat patches were created to simulate various conditions of rainforest restoration which may be experienced by colonising arthropods (Figure 1.1). In order to test for the impacts of a glyphosate herbicide (a commonly used herbicide in rainforest restoration) on the composition of arthropod assemblages, I carried out a separate field experiment in which experimental quadrats were established within intact rainforest. The experiment aimed to reduce and control the great diversity and variation inherent in pre-existing restoration plots (the usual subjects of such studies).

Before the field experiments, a survey was carried out to collect information about the distribution of soil and litter arthropods in remnant rainforests (undisturbed reference sites) and cleared pasture (disturbed reference sites) of the study region. The information obtained from the survey was used to a) generate potential bio-indicators of forested and cleared habitats and b) evaluate the utility of different taxonomic groups and sampling methods. The information obtained from the survey also served as reference data that were utilized in the analyses of the experimental results.

This thesis is divided into eight chapters (Figure 1.1). Chapter 2 gives an overview of the project methodology with particular emphasis on the manipulative field experiment (Chapters 4-6). Chapter 3 deals with the baseline survey. Chapters 4, 5 and 6 describe and discuss the manipulative field experiment: Chapter 4 investigates the effect of isolation and the efficacy of inoculation; Chapter 5 tests the effects of quality and quantity of mulch that is used during the initial stages of restoration; and Chapter 6 investigates the effects of shading and litter (mulch) depth with implications for different types of restoration. Chapter 7 describes the additional field experiment that tested the impacts of herbicide on soil and litter arthropod assemblages in the floor of rainforest. The concluding Chapter 8 summarises the findings of each of the main chapters (Chapters 3-7) and presents general discussions and recommendations for restoration practitioners.

Chapters 3 through 7 are presented here as manuscripts for publication in peer-reviewed journals. Although individual chapters address the separate issues listed above, some repetitive statements (especially within Methods sections) inevitably appear.

24 Manipulative field experiment (Chapters 4-6) Baseline survey (Chapter 3) Manipulative field experiment I: Isolation and inoculation • Assesses impacts of (Chapter 4) rainforest clearing and subsequent pasture use on • Investigates the effect of isolation on the soil and litter arthropods. patterns of colonisation of restored • Identifies bio-indicators of habitat patches in a matrix of cleared rainforest and pasture pasture by assemblages of soil and litter habitats. arthropods. • Evaluates utilities of • Tests efficacy of inoculation, which different taxonomic groups involved translocation of litter and sampling methods. (containing live arthropods) from • Provides baseline rainforest habitat to isolated habitat information for the field patches. experiment. Manipulative field experiment II: Mulch quality and quantity (Chapter 5)

• Assesses effects of mulch quality (hay vs woodchip mulch) and quantity (shallow vs deep) on the development of arthropod assemblages in experimentally created habitat patches that simulated newly replanted restoration. Herbicide field experiment Manipulative field experiment III: (Chapter 7) Shading and mulch depth (Chapter 6) • Assesses immediate and delayed impacts of • Assesses effects of different levels of glyphosate herbicide shading (0, 50, 90%), and depth of application on rainforest soil woodchip mulch (shallow vs. deep) on and litter arthropods. the development of arthropod assemblages.

Figure 1.1 Conceptual flow chart of the thesis data-chapters, explaining aims and interrelationships with other data-chapters (indicated by arrows).

25

26 2 OVERVIEW OF METHODOLOGY

2.1 Study area

The study area for this project was located on the Maleny plateau, in the Sunshine Coast hinterland of eastern Australia, ca. 100 km north of Brisbane (26°40’- 50’ S, 152°45’- 53’ E, elevation 250 to 530 m) (Figure 2.1). This region was selected because local information (i.e. contact details of land owners, land use history) was readily accessible through a local community group (Barung Landcare), and there were sufficient rainforest remnants and mature rainforest regrowth areas (age of ca. 100 years) to obtain the required number of study sites.

Legend

Major road - 0 5 km ― Creek Baseline survey site X Experiment site

Figure 2.1 A map of the study area, showing 12 paired sites used in the baseline survey (Chapter 3) and five experimental sites used in Chapters 4-7. Each experimental site comprised a combination of one main (E1-5) and one ‘distant’ sub-site (D1-5) (see text for more details).

27 There was an extensive coverage of rainforests over the entire region of Maleny plateau before European settlement in 1875 (Gubby 1994). Following this, most of the valuable timber trees, such as Red Cedar (Toona ciliata), were logged within a few decades, and extensive areas of subtropical rainforest were cleared for dairy farming up to the 1930’s (Maleny and District Centenary Committee 1978; Gubby 1994). Regrowth of rainforest occurred in some locations where dairy productivity was low and farming subsequently abandoned, mostly in small areas and often on steep slopes. From around the 1980s to the time of the present study (2003-2004), the intrinsic value of rainforest in the region received an increasing level of official recognition, and government funds were provided to support rainforest restoration in the Maleny region from the mid-1990s (Catterall & Harrison 2006).

The major land uses in the Maleny region are pasture, macadamia and timber plantations, sclerophyll forest, and subtropical rainforest. At the time of the present study, large parts of the Maleny region were used for dairy or beef farming, or were being transformed into residential areas. Kikuyu grass (Pennisetum clandestinum), exotic to Australia, is widely used for grazing; however, other less common pasture species such as Rhodes grass (Chloris gayana), Paspalum spp. and clover (Trifolium spp.), were also common in grazed pastures. Rainforest sites included in this project were either old regrowth (age of ca.100 years) or remnant forest that had been selectively logged until the mid to late 1900s. Structural types (Adam 1992) of the rainforest sites were either complex notophyll vine forest, or notophyll feather palm vine forest (see also Plate 1). Soils of most pasture and rainforest sites used in this project were basaltic. At least some of the areas, located on the escarpment edge had a combination of basaltic soil and colluvium or metamorphic rocks (see Appendix 2).

Mean daily maximum and minimum temperatures in mid summer (January) are 28.9 and 18.8 ° C respectively, and 19.5 and 7.1 ° C in mid winter (July). Frost is common in winter (June-August). Average annual rainfall in the region is 1851 mm, with most (170-310 mm per month) falling between December and April (climate data averaged over at least 50 years up to 2004 obtained from the Australian Bureau of Meteorology). The total precipitation during the entire study period (2917 mm between January 2003 and May 2004) was slightly below the expected average rainfall (3186 mm).

28 Detailed characteristics of each study site are described in relevant chapters and Appendix 2.

2.2 Study design

2.2.1 Components of the study

The research described in this thesis consists of three components: a baseline survey of arthropods in pasture and rainforest (Chapter 3), a manipulative field experiment to test for effects of restoration methods on colonising arthropods (Chapters 4-6), and a field experiment to test for the impacts of herbicide application on arthropods of the rainforest floor (Chapter 7). Detailed description of the study designs were provided in their respective chapters (Chapters 3-7). The remainder of this chapter gives an overview of the manipulative field experiment and describe related pilot studies that are not thoroughly dealt in Chapters 4 to 6.

2.2.2 Manipulative field experiment

The manipulative field experiment was designed to test the effects of several factors on the development of assemblages of soil and litter arthropods potentially colonising from surrounding habitats. Small-scale habitat patches were created to simulate various environmental conditions that may be experienced by soil and litter arthropods during rainforest restoration on former agricultural land.

A total of five experimental sites, each comprising a combination of one main (E1-5) and one ‘distant’ sub-site (D1-5) (Figure 2.1), were dispersed across a study region of around 170 km2; all within 13 km of the township of Maleny. Each site consisted of cleared landscape (pasture) abutting a rainforest remnant. The size of the rainforest remnants varied from 1.15 ha at one site (E2, D2), to over 10 ha at other sites (see also Appendix 2). The pasture was grazed by cattle and/or horses. In all cases, the pasture grasses formed a dense layer, completely covering the soil surface.

At each of three sites (E2, D2; E4, D4; E5, D5), main and ‘distant’ sub-sites were located in the same area, sharing the same rainforest remnants. This was not the case for the two other sites (E1, D1; E3, D3), as no suitable location could be found to establish both main and ‘distant’ sub-sites within the same area (Figure 2.1). Although three experimental

29 sites (E2, D2; E4, D4; E5, D5) were located relatively close to each other (1-2.5 km away), these sites did not share the same remnant rainforest.

At each site, I established a series of 14 experimental plots (3 x 3 m quadrats) that simulated the environmental conditions experienced by soil and litter arthropods within areas of replanted rainforest. Experimental plots were placed at 0 (9 plots), 15 (1), 100 (1) and 400-500 m (2) from the rainforest edge (see also Plate 3). One of the plots at 400-500 m from the rainforest edge was established to test the efficacy of inoculation (Table 2.1), which involved translocation of litter (containing live arthropods) from within the rainforest remnants (see Chapter 4 for more details of inoculation). A control plot was established within each rainforest remnant, between 30 m and approximately 100 m inside the rainforest edge. The nine plots at the edge of the rainforest (i.e. 0 m) were located in pasture at approximately two metres outside the edge of the rainforest undergrowth or canopy cover, whichever was closer. The plots were established at least five metres apart from each other to maintain some independence (Figure 2.2).

Based on discussions with revegetation practitioners in the study region, experimental plots were designed to create environmental conditions which occur as a result of actual replanting procedures (Kanowski et al. 2003; Big Scrub Rainforest Landcare Group 2005; Catterall et al. 2007). First, broad spectrum herbicide (Biactive Round Up®, 7.2 g/L Glyphosate) was sprayed over each plot to kill all existing vegetation. Three weeks after the application, all visible vegetation (dead and alive) was removed by hand. Steel posts, 1.2 m high were then erected at each corner of the plot which was fenced with barbed wire to exclude stock (Figure 2.3).

To simulate shading effects produced by a developing tree canopy, the plots were either unshaded or covered with Sarlon® shadecloth rated at either 50% or 90% protection from insolation. Shadecloth was placed over the top of the entire quadrat (3 x 3 m, at a hight of around 1.2 m), and 20-40 cm down each side. The shadecloth was cut in 15-20 places with a slit length of 10-15 cm to permit sunflecks and throughfall of rain.

30 Table 2.1 Description of experimental plots established at each site (see also Figure 2.2). ‘Woodchip’ or hay mulch was placed at a depth of 10-15 cm (‘deep’), or 3-5 cm (‘shallow’). Shadecloth rated at either 50% or 90% protection from insolation was used to cover some of the experimental plots. See text for more details. Plots used for

analyses†

Distance Plot no. from the (see Fig. rainforest Mulch 2.2) edge Shade Mulch type depth Note Ch.4 (Isolation) Ch.4 (Inoculation) Ch.5 Ch.6 1 400-500 m 90% Woodchip Deep Y Y 2 400-500 m 90% Woodchip Deep Inoculated Y 3 100 m 90% Woodchip Deep Y 4 15 m 90% WoodchipDeep Y 5 0 m 90% WoodchipDeep Y Y 6 0 m 90% WoodchipShallow Y 7 0 m 50% WoodchipDeep Y 8 0 m 50% WoodchipShallow Y 9 0 m 0% WoodchipDeep YY 10 0 m 0% WoodchipShallow YY 11 0 m 0% Hay Deep Y 12 0 m 0% Hay Shallow Y 13 0 m 0% n/a n/a Un-mulched control YY 14 -30 m 90% Woodchip Deep Rainforest control Y † Plots with ‘Y’ symbol were used for data analyses in the corresponding chapter.

To provide a potential habitat for soil and litter arthropods, a mulch of woodchip and leaf material (referred to as ‘woodchip mulch’ hereafter) or hay was placed at two different depths (‘shallow’, 3-5 cm; ‘deep’, 10-15 cm) in a square area (2.5 x 2.5 m) established inside each plot (Figure 2.3). The area was bordered by wire netting (40 cm high, hexagonal mesh size of 1.5 cm [maximum height] x 2 cm [maximum width]) to minimise loss of mulch due to wind or disturbance by wildlife. Woodchip mulch was derived from lopping vegetation around powerlines in various locations within about 150 km of the study region, conducted by a local contractor of the electricity provider (Smokey Flats Timber Cutting Pty. Ltd., Wacol, Queensland). Woodchip mulch comprised a mix of foliage and wood, derived from rainforest and eucalypt species. Hay was obtained from a nursery (‘The Hay Man Nursery’) in Peachester, south of Maleny. The main constituent

31 of the hay was millet (Panicum spp.), one of the commonly used hay mulches in rainforest plantings in the region.

‘Distant’ sub- site (D1-5, see

Fig. 2.1 for their Main sub-site (E1-5, see Fig. 2.1 for locations) their locations)

Rainforest Pasture remnant

400-500 m 100 m15 m 0 m – 30 m Patch of trees (<1 ha) (1)

(3) 50 m (min.) (4) (14) 80m (min.) (2) (5) Deep woodchip mulch 5-10 m 250 m (min) (6) Shallow woodchip mulch (7) Deep woodchip mulch

(8) Shallow woodchip mulch (9) Deep woodchip mulch

(10) Shallow woodchip mulch Sclero phyll forest (11) Deep hay mulch

(12) Shallow hay mulch Legend

Plot with 90% shadecloth (13) None Plot with 50% shadecloth Plot with no shadecloth (0%) Inoculated plot (with 90% shadecloth)

Figure 2.2 Schematic diagram of one of the field experimental sites (distances not to scale), showing criteria used when positioning experimental plots. ‘Deep’ indicates sterilised ‘woodchip’ or hay mulch placed at a depth of 10-15 cm, and ‘shallow’ is 3-5 cm. The plots 1-4 and 14 contained deep woodchip mulch. The nine plots along the edge were located randomly. Plot numbers in parentheses corresponds to those in Table 2.1. See also Plates 3, 5-8.

32 Before mulch was added to the experimental plots, it was sterilised by steam-treatment using the following method. A landscaping dump truck was modified to fit steam pipes on the floor of the open-top steel container (volume of approximately 7m3). Steam pipes consisted of steel pipes (approximately 4 cm in diameter) with a number of small steam vents along each. These pipes were arrayed parallel (ca. 50 cm apart from each other) on the floor of the container and were attached to a diesel-powered steam generator (operated by Steam Services Pty. Ltd. in Wacol, owned by the Queensland Department of Primary Industries Fire Ant Research Group). Woodchip mulch or hay was loaded into the container with its top covered with tarpaulin sheet, and then steam-treated (at a minimum of 95 C°) for 100 minutes. A total of approximately 35 m3 of woodchip mulch and 5 m3 of hay were steam-treated and used for the entire field experiment (see also Plate 4).

A control plot was constructed inside the rainforest remnant at each site in exactly the same manner as other plots except that a tent of plastic bird netting (mesh size of 2 x 2 cm) covered the entire plot, minimising the natural fall of rainforest leaves and twigs into the experimental area (plot no. 14, see Table 2.1 and Figure 2.2, see also Plate 5a, b). Another control plot was also constructed along the rainforest/pasture edge (0 m) using the same methods and materials except that it was not covered by shadecloth, and received no mulch (plot no. 13, shown in Table 2.1 and Figure 2.2, see also Plate 5c).

┼ ┼

■ ■

x x 80 cm Legend ● Plot boundary (fenced with barbed wire) x x Mulch boundary (fenced with wire netting) ● Centre of the plot (with a 1.2 m steel ■ ■ post) X Pitfall trap 2.5 m ■ Timber stake (40 cm high) ┼ Steel post (1.2 m high) ┼ ┼

3 m Figure 2.3 Layout of an experimental plot (not to scale). When present, shadecloth covered the area within the solid line.

33 2.2.3 Pilot studies

Mulch steaming trials To ensure the efficacy of mulch sterilisation, the duration of steaming required to kill all the existing arthropods in woodchip and hay mulch was discussed among the staff at the steaming facilities, and a duration of 40 minutes was initially proposed. A trial run was carried out by steaming 7 m3 of woodchip mulch. Immediately after the completion of the trial steaming, small quantities of mulch (approximately 20 L) were collected from the bottom, middle and top of the batch in the steaming container. In order to test for the presence of live arthropods, the treated mulch was processed with Tullgren funnels. A piece of moistened cardboard was placed on the bottom of the collection vials to keep any extracted arthropods alive. The resulting extracts were checked every 4-8 hours for a total of 5 days. It was found that 40 minutes of steaming was insufficient to sterilise the mulch as some live arthropods (e.g. mites and pseudoscorpions) were found. Consequently this procedure was repeated for a steaming duration of 100 minutes and no live arthropods were observed in the subsequent Tullgren extracts.

Field experiment pilot study Before beginning the full-scale manipulative field experiment, a pilot experiment was carried out to test the feasibility of the experimental design. One 3.5 x 14 m plot (in pasture) and one 3.5 x 5 m plot (within remnant rainforest) were established at one of the experimental sites (E1, see also Figure 2.1). Construction of the pilot plots commenced on 19 September and was finished on 11 November 2002. The plots were left at the sites until April 2003 to test their durability and to estimate the frequency of visitation required for adequate maintenance of the plots. Arthropod samples were not collected from these plots, as the initial mulch steaming (40 minutes duration) used for the pilot study was found to be insufficient to kill all of the existing arthropods.

The study design was extensively modified after the pilot study. Major amendments included reduction of the plot size to 3 x 3m. This reduced the construction cost per plot, allowing an increase in the number of the plots from 20 to 70 with additional experimental treatments and replications. In addition, timber stakes were replaced with steel poles (erected at corners of the plots, see Figure 2.3) as they were cheaper and easier to manipulate.

34 2.3 Sampling methodology Arthropod sampling was carried out using pitfall traps and extraction from litter and surface soil. Pitfall trapping samples the relative density of active epigaeic arthropods, whereas litter extraction samples the relative abundance of sedentary arthropods that live in litter and topsoil (Majer 1997). Both sampling methods are commonly used for collecting soil- and litter-dwelling arthropods. Pitfall traps are expected to become more effective as the amount of litter decreases, whereas litter extraction is more effective in closed forest with ample litter. These two sampling protocols were, therefore, expected to complement each other in habitats (or experimental plots) with different amounts of litter in the present study. For example, in rainforest plots with heavy litter cover, litter extraction could be the more effective method, while in pasture plots with poor litter coverage, pitfall trapping could be more effective. Both sampling protocols were employed in the experiments presented in the all following chapters except for the herbicide field experiment (Table 2.2) in which only litter extraction was employed (see Chapter 7).

Table 2.2 Sampling methods and taxonomic resolutions of arthropods used in respective chapters (indicated by ‘Y’ symbols). Sampling Arthropod groups used method for analysis

Chapter Pitfall traps traps Pitfall extraction Litter ‘Coarse’ arthropods Ant species Ant genera Ant functional groups Ant biogeographical affinities Ch.3 Baseline survey of arthropods in Y Y Y Y Y Y Y pasture and rainforest Ch.4 Manipulative field experiment I: Y Y Y Y Y isolation and inoculation Ch.5 Manipulative field experiment II: Y Y Y Y Y mulch quality and quantity Ch.6 Manipulative field experiment III: Y Y Y Y Y shading and mulch depth Ch.7 Herbicide field experiment Y Y Y

35 All insects and arachnids were sorted to Order (except Hymenoptera which were split into Formicidae and ‘other Hymenoptera’) and myriapods to Class. Collembola, Acari and Diptera (with the exception of litter-extracted dipterans) were not sorted due to their high abundance and ubiquitous occurrence regardless of habitat types. The dataset consisting of arthropods sorted at coarse taxonomic resolutions is hereafter referred to as ‘coarse’ arthropods. Ants (Hymenoptera: Formicidae) were selected as a target group, and identified to genus and then sorted to species. With assistance from Alan Andersen (CSIRO Sustainable Ecosystems), ant species were also assigned to functional groups (Andersen 1995a) and their biogeographical affinity. Although many groups of arthropods live in soil during at least one stage of their life cycle, particularly during their larval stage ('part-time inhabitants' sensu Decaens et al. 2006), the present study focuses only on adult arthropods. Voucher specimens are deposited at Queensland Museum and Griffith School of Environment, Griffith University.

Coarse arthropod and ant species datasets were used in the analyses of all research components (Table 2.2). The ant genus dataset was used only in the baseline survey. The ant functional group dataset was used in all analyses except the herbicide field experiment, as functional groups were not expected to respond differently to ecotoxicological effects of the herbicide.

Detailed descriptions of sampling protocols, sorting and data analyses are included within the following chapters.

36 3 BASELINE SURVEY

3.1 Introduction

Forest clearance, mainly for agricultural development, is widely recognised as the most serious anthropogenic threat to biodiversity (Sala et al. 2000). The world’s forest area was reduced by 2.4 per cent in the 1990’s (United Nations Department of Economic and Social Affairs 2002). In Australia, large tracts of land have been cleared since European settlement (Greenslade & New 1991). A large number of studies have reported declines in vertebrate and some charismatic arthropod (e.g. butterflies) species in response to clearing of forest (Watt et al. 1997; Myers et al. 2000; Sodhi et al. 2004). However, little attention has been paid to other organisms (but see Basset et al. 1998; Lewinsohn et al. 2005a). Since no particular taxon can necessarily be a surrogate for another (Lawton et al. 1998; Alonso 2000), studies are needed to investigate the impacts of deforestation on less well-known taxa such as soil and litter arthropods.

Despite their inconspicuous nature, soil and litter arthropods are one of the most prominent components of ecological communities in terms of both abundance (biomass) and diversity (Ghilarov 1977; Andre et al. 1994). Soil and litter arthropods are also known to interact with, or mediate many ecological processes, including the decomposition of organic matter (Reichle 1977; Petal 1978).

In addition to their ecological importance, the use of terrestrial arthropods as bio-indicators of habitat change has been suggested by a number of authors (e.g. Kremen et al. 1993; Andersen & Majer 2004). If we can quantify the effects of clearing on soil and litter arthropods we can develop bio-indicators of cleared and forested habitats that may be applied to intermediate habitat types resulting from both forest degradation (e.g. partial clearing or fragmentation) and forest restoration (e.g. regrowth or replanting). These bio-indicators, based on the composition of soil and litter arthropod assemblages, should measure whether degradation is occurring or whether restoration is successful. Establishing reliable bio-indicators, and their characteristic values in reference areas such as forest and cleared areas, has often been neglected in studies of degradation or restoration.

37 In this chapter I assess the impacts of rainforest clearance and associated land use for pasture, on assemblages of soil and litter arthropods. I compare sensitivities to changes in forest cover at different taxonomic levels. At lower taxonomic and ecological levels (species, genus and functional group) the study focuses on ants. These results are, therefore, relevant to the current debate on ‘taxonomic sufficiency’ (Pik et al. 1999). I identify potential bio-indicators of rainforest and pasture that may be used within programs that monitor rainforest degradation or restoration. In addition, I compare the performance of two commonly-used sampling methods: pitfall trapping and litter extraction.

3.2 Methods

Study area The study was undertaken on the Maleny plateau, in the Sunshine Coast hinterland of eastern Australia, ca. 100 km north of Brisbane (26°40’- 50’ S, 152°45’- 53’ E, elevation 250 to 530 m). Following European settlement in Maleny in 1875, extensive areas of subtropical rainforest were cleared for dairy farming up to the 1930’s (Gubby 1994). Large parts of the Maleny plateau are currently used for dairy or beef farming, or are being transformed into residential areas.

Mean daily maximum and minimum temperatures in mid summer (January) are 28.9 and 18.8 ° C respectively, and 19.5 and 7.1 ° C in mid winter (July). Frost is common in winter (June-August). Average annual rainfall in the region is 1851 mm, with most (170-310 mm per month) falling between December and April.

Twenty-four sites were selected within 13 km of the township of Maleny; 12 in rainforest remnants and 12 in pasture (N =12). Rainforest and pasture sites were paired at each location (see also Figure 2.1). Most pasture and rainforest sites were on basaltic soil. At five sites located on the escarpment edge, a combination of basaltic soil and colluvium or metamorphic rocks was present (Appendix 2).

Rainforest sites were either old regrowth (age of ca.100 years) or remnant forest that had been selectively logged until recently. Structural types (Adam 1992) of the rainforest sites were either complex notophyll vine forest, or (for two sites located on the escarpment of the plateau) notophyll feather palm vine forest (Appendix 2). Ten of the 12 pasture sites

38 were heavily grazed by cattle and/or horses. One pasture site was lightly grazed, and another site was unstocked but was occasionally machine-mown (Appendix 2). Kikuyu grass (Pennisetum clandestinum), exotic to Australia, has been extensively sown at many sites; however, other less common pasture species such as Chloris gayana, Paspalum spp. and Trifolium spp. were also present. In all cases, the pasture grasses formed a dense layer, completely covering the soil surface.

Sampling methodology Sampling was carried out between early January and early May 2003, the warmer and wetter seasons, when arthropod abundance was expected to be relatively high (Andersen 1986; Frith & Frith 1990).

At each site a transect of ca. 60 m, running approximately parallel to the edge of each habitat type, was established. At a number of rainforest sites the transect was reduced to ca. 50 to 30 m due to the size and shape of the remnant. Along each transect, three sampling points were positioned 15 to 30 m apart. Each point consisted of a circular area of 3 m radius. In rainforest remnants, all sampling points were at least 30 m from the rainforest edge except for one which was 22 m from the edge (see also Appendix 2). In the pasture sites, all sampling points were at least 100 m away from the rainforest remnants.

Four pitfall traps were set at distances of 1 to 2 m around the centre of each sampling point, evenly spaced (i.e. 12 traps per transect). Each pitfall trap was a 120 ml plastic vial (44 mm in diameter), buried in the ground with the lip flush with the surface. Vials were left unused with the lid on for at least a week after installation to avoid ‘digging-in’ effects (Greenslade 1973). Vials were then filled with 70 to 80 ml of 70% ethanol: a small amount of glycerol was added to reduce evaporation. Pitfall traps were left open for five days. The twelve pitfall-trapped samples from each site were pooled for data analysis.

Litter extraction was carried out by collecting 1 litre of litter and surface soil (to a depth of 1 to 2 cm) in small amounts evenly over the entire circular area at each sampling point (i.e. 3 collections per transect). The same proportion of surface soil relative to litter was collected in all of the sampling points regardless of the habitat type (approximately 20% surface soil and 80% litter by volume). Each soil and litter sample was placed in a fabric bag and kept in an insulated box. Samples were placed in Tullgren funnels within 12

39 hours of sampling. Arthropods were extracted for 4.5 days using 40 W clear light bulbs. The three extracted samples from each site were pooled for data analysis.

All insects and arachnids were sorted to Order (except Hymenoptera which were split into Formicidae and ‘other Hymenoptera’) and myriapods to Class. Collembola, Acari and Diptera (with the exception of litter-extracted dipterans) were not sorted due to their high abundance and ubiquitous occurrence regardless of habitat types. All individuals smaller than 1 mm were excluded from pitfall samples as they tended to be overlooked in samples that were contaminated with soil. Ants (Formicidae) were selected as a target taxon, and identified to genus and sorted to species. Where possible, ants were identified as described species using published taxonomic literature, otherwise they were assigned species codes. With assistance from Alan Andersen (CSIRO Sustainable Ecosystems), ant species were allocated to functional groups (see Table 1.1) and groups reflecting their biogeographical affinities (viz. Bassian, Eyrean, Torresian, Widespread; see Andersen 2000b).

Data analysis My study was a balanced pair-wise (pasture and rainforest remnant) design amenable to both univariate and multivariate analysis. The unit of replication was the transect (N = 12 pasture and rainforest transects), at 12 sites. Data were analysed as five different datasets: arthropods sorted to orders/class (referred as ‘coarse arthropods’ hereafter), ant genera, ant species, ant functional groups and ants grouped into their biogeographical affinity. Each dataset contained information obtained from two different sampling methods (pitfall trapping and litter extraction). Abundances (total individuals collected at a site from either 12 pitfall traps or three litter collections) of coarse arthropods were log transformed before analysis. Data on the genera and species of ants were expressed as frequency scores of 0 to 3 (see Nakamura et al. 2003 for more details). This was done as ant abundances vary considerably depending on the proximity of sampling points to colonies (Andersen 1991b). Abundances within ant functional groups and biogeographical groups were expressed as proportions of all ants at each site, so that the dominance by different groups could be compared.

Multivariate analyses were carried out to investigate differences in the structure of arthropod assemblages between pasture and rainforest habitats using non-metric multi-scaling ordination (NMDS) executed using PRIMER v.5 software (Clarke 1993).

40 All NMDS ordinations were performed on Bray-Curtis similarity matrices, with 10 random restarts. Analysis of Similarity (ANOSIM: Clarke 1993), with 999 permutations, was used to test the significance of differences in assemblage composition between pasture and rainforest. Test statistics (Global R values) derived from ANOSIM, which represent measures of the degree of separation, were used to compare relative discriminatory powers of the various datasets. ANOSIM was also carried out on combined data from pitfall traps and litter extraction, range-standardised across each taxon, to investigate the similarity of patterns in assemblage composition between the two methods.

Finally, coarse arthropods, ant genera, ant species, ant functional groups and biogeographical affinities that were characteristic of either rainforest or pasture habitats were identified using the Indicator Value method of Dufrene & Legendre (1997). This method combines measures of fidelity (relative frequency) and specificity (relative abundance) to generate indicator values (IndVals) of each taxon expressed as percentages. A random reallocation procedure with 5000 permutations was used to test for the significance of IndVals. When a taxon was found to be a statistically significant habitat indicator, that taxon was classed as either a habitat ‘specialist’ (when it was found exclusively in its preferred habitat) or habitat ‘increaser’ (when it was found in both habitat types but occurred primarily in their preferred habitat). I used a subjective benchmark value for IndVal of > 70%, as proposed by McGeogh et al. (2002), as defining characteristic indicators for the habitats in question. Parallel t-tests were also carried out to test for differences in abundance (or frequency scores for ants) between rainforest and pasture.

As many arthropods are known to have patchy spatial distributions (Giller 1996; Soares & Schoereder 2001), their reliability as habitat indicators may be low when individual taxa are analysed separately. To develop a more robust indicator statistic for potential application to disturbed or revegetated sites, I used the following procedure for the selected datasets (viz. coarse arthropods, ant genera, ant species). Abundances of all statistically significant taxa within the coarse arthropod dataset for a particular habitat (rainforest or pasture) were individually range-standardised (site-specific abundance minus minimum abundance across all sites / maximum minus minimum) to give values between 0 and 1 for each taxon at each site. Range-standardisation was not carried out for ant genera and species, as each taxon had been given a frequency score of 0 to 3. These 41 abundance values were then summed across all statistically significant indicator taxa at each site, giving a single value which quantifies the extent to which a site is rainforest-like or pasture-like, in terms of its arthropod assemblage. The utility of these ‘composite indices’ was then verified by using them to calculate new IndVals, whose statistical significance was tested.

Although multiple statistical testing was involved in this study, no Bonferroni corrections were employed, and the significant P value was set at 0.05. This is because avoidance of Type II error was important for the purpose of this study, designed to screen for taxa which responded to habitat change (Roback & Askins 2005, see also Moran 2003).

3.3 Results

Assemblage composition A total of 14555 arthropods were sampled (8988 from pitfall traps and 5567 from litter extraction). Ants were the most abundant taxon with 4236 individuals, followed by Coleoptera with 2355 and ‘Homoptera’ (Auchenorrhyncha plus Sternorrhyncha) with 1080. Thirty coarse arthropod taxa were identified. Many were represented by more than 100 individuals in total and occurred at numerous sites (see also Appendix 3). Forty-four ant genera and 84 ant species were identified, but unlike coarse arthropod taxa, their occurrence was patchy. Nine ant functional groups (Cryptic species, Cold climate specialists, Dominant Dolichoderinae, Generalised Myrmicinae, Hot climate specialists, Opportunists, Subordinate Camponotini, Specialist predators, Tropical climate specialists) and four groups of biogeographical affinities (Bassian, Eyrean, Torresian, Widespread) were recorded.

NMDS ordinations (Figure 3.1) and ANOSIM Global R values (Table 3.1) showed that most datasets (coarse arthropods, ant genera, species and functional groups) separated rainforest from pasture sites. According to the Global R values, ant species best segregated rainforest from pasture. Coarse arthropods exhibited the second highest Global R values followed by ant genera and functional groups. Only coarse arthropods from pitfall traps detected differences in grazing intensity; two pasture sites where grazing intensity was conspicuously low fell within the rainforest cluster on the ordination (Figure 3.1a). Ant biogeographical affinity failed to show any patterns between pasture and rainforest (Pitfall traps, Global R = -0.019, P = 0.574; Litter

42 extraction, Global R = -0.032, P = 0.693; Methods combined, Global R = -0.049, P = 0.825).

a Coarse arthropods b Ant genera Stress: 0.17 Stress: 0.22

Stress: 0.17 Stress: 0.22

c Ant species d Ant functional groups

Stress: 0.16 Stress: 0.14

Stress: 0.16 Stress: 0.14

□ Pasture, Pitfall ■ Rainforest, Pitfall ◇ Pasture, Litter extraction ◆ Rainforest, Litter extraction

Figure 3.1 NMDS ordination of rainforest and pasture sites based on (a) coarse arthropods, (b) ant genera, (c) ant species and (d) ant functional groups. Points sampled using pitfall traps and litter extraction are shown separately. Note that two sites with low grazing intensity are indicated by crossed boxes.

At one rainforest site, invasive coastal brown ants (Pheidole megacephala) contributed more than 75 percent of the total ant catch from pitfall traps (see also Appendix 3b). However this site still fell within the rainforest cluster on the ordination, as other ants within the same site were characteristic rainforest species. Pheidole megacephala was not found at any other rainforest or pasture site.

43 Table 3.1 Global R values of each dataset comprising assemblages of rainforest and pasture, using different sampling methods. Note that all P values for Global R were < 0.001. Dataset Sampling method Global R Coarse arthropods Pitfall traps 0.6331 Litter extraction 0.801 Methods combined 0.850

Ant genera Pitfall traps 0.477 Litter extraction 0.649 Methods combined 0.698

Ant species Pitfall traps 0.831 Litter extraction 0.812 Methods combined 0.934

Ant functional groups Pitfall traps 0.587 Litter extraction 0.285 Methods combined 0.572 1Global R without samples collected from two low grazing intensity sites: 0.751 (see also Figure 3.1).

Indicators of rainforest and pasture At all taxonomic resolutions, the most statistically significant indicator taxa were of rainforest, with fewer indicative of pasture (Tables 3.2-3.5). Parallel t-tests showed that most taxa with significant IndVals were also significant in the t-tests, suggesting their abundances (or frequency scores for ants) were higher in their associated habitats.

Among coarse arthropods (Table 3.2), a number of taxa (viz. Pseudoscorpionida, Pauropoda, Isopoda, Blattodea, Diplopoda, Chilopoda, Homoptera) were found to be outstanding rainforest or pasture indicators with their IndVals exceeding the cut-off values of 70 percent based on pitfall trapping and/or litter extraction. The majority of them were habitat ‘increasers’. Only two taxa (viz. Archaeognatha, Opilionida) were ‘specialists’; however, both occurred patchily, and their IndVals were both under 70 percent.

Compared with coarse arthropods, fewer ant genera were found to be significant habitat indicators, and most exhibited IndVals of less than 70 percent, suggesting many were not strong indicators of habitat (Table 3.3). Among the significant indicators, only Leptomyrmex, Hypoponera and Paratrechina exceeded the cut-off values of 70 percent, based on either pitfall trapping or litter extraction. 44 Table 3.2 Mean abundances and Indicator values (IndVals) of coarse arthropod habitat indicators. Only statistically significant indicators for each sampling method are shown. Pitfall traps Litter extraction RM1 PM1 IndVal2 Type3 RM1 PM1 IndVal2 Type3 Rainforest indicators Rainforest indicators Archaeognatha 1.3 0 66.7** S Opilionida 1.4 0 58.3* S Opilionida 1.1 0 58.3* S Pseudoscorpionida 18.3 0.1 97.9** I Pseudoscorpionida 5.0 0 83.3** I Pauropoda 9.5 0.1 88.6** I Blattodea 4.6 0.5 82.4** I Isopoda 34.8 3.1 87.5** I Diplopoda 8.8 1.1 82.4** I Diplopoda 25.8 1.5 86.1** I Isopoda 8.3 6.6 67.7* I Chilopoda 8.4 2.3 73.6** I Dermaptera 6.8 1.7 67.5* I Coleoptera 46.9 12.5 64.5** I Diplura 6.0 1.3 66.0* I Heteroptera 21.8 4.3 63.8* I Heteroptera 28.1 15.4 62.2* I Other Hymenoptera 4.4 2.0 63.1 I Coleoptera 97.4 39.4 58.7* I Diplura 15.6 1.7 63.0 I Psocoptera 1.4 0 41.7 I Symphyla 6.5 0.2 62.3* I Formicidae 79.3 36.7 56.4* I Blattodea 0.8 0.1 52.2 I Dermaptera 0.9 0.3 52.2 I Amphipoda 1.5 0.3 39.9 I Pasture indicators Pasture indicators Homoptera 3.1 35.0 72.3** I Orthoptera 0.1 1.3 45.7* I Orthoptera 5.3 44.0 66.4** I Araneae 9.8 45.3 61.9* I 1RM, PM, mean abundances in rainforest and pasture respectively. 2Indicator values (IndVals) and the results of their permutation test: * = P < 0.01, ** = P < 0.001, otherwise P < 0.05. 3S, I, habitat ‘specialists’ and ‘increasers’ respectively (see the Methods for more detail).

Table 3.3 Mean frequency scores and Indicator values (IndVals) of ant genus habitat indicators. Only statistically significant indicators for each sampling method are shown. Pitfall traps Litter extraction RM1 PM1 IndVal2 Type3 RM1 PM1 IndVal2 Type3 Rainforest indicators Rainforest indicators Leptomyrmex 1.00 0 75.0** S Lordomyrma 0.50 0 41.7 S Anonychomyrma 0.83 0 58.3* S Hypoponera 2.50 0.42 85.7** I Notostigma 0.67 0 50.0* S Monomorium 1.08 0 66.7** I Prolasius 0.50 0 33.3 S Strumigenys 0.83 0 58.3* I Pheidole 2.67 1.58 62.7 I Mayriella 0.75 0 50.0* I Leptogenys 1.25 0.33 59.2* I

Pasture indicators Pasture indicators Cardiocondyla 0 0.58 50.0* S Paratrechina 0 1.50 75.0** I Polyrhachis 0 0.33 33.3 S Paratrechina 0.50 1.83 52.4 I Carebara 0.17 0.83 41.7 I 1RM, PM, mean frequency scores in rainforest and pasture respectively. 2Indicator values (IndVals) and the results of their permutation test: * = P < 0.01, ** = P < 0.001, otherwise P < 0.05. 3S, I, habitat ‘specialists’ and ‘increasers’ respectively (see the Methods for more detail). 45 When ants were analysed at species level (Table 3.4), the majority of indicator species were habitat ‘specialists’. In contrast, coarse arthropods had more habitat ‘increasers’. Amongst habitat indicators showing IndVals of over 70 percent, only Hypoponera sp.1 was a habitat ‘increaser’; the others (viz. Pheidole QM2, Pheidole sp.2, Rhytidoponera metallica, Paratrechina QM2) were habitat ‘specialists’. Although significant, the occurrences of most remaining indicator species were rather sporadic, exhibiting IndVals of less than or equal to 50 percent.

For ant functional groups (Table 3.5), all of the statistically significant indicator groups were found to be habitat ‘increasers’. Specialist predators, Generalised Myrmicinae, Tropical climate specialists and Cold climate specialists were statistically significant rainforest ‘increasers’, with IndVals of over 70 percent in pitfall traps. Opportunists were strong pasture ‘increasers’ in both trapping methods. None of the IndVals based on ant biogeographical affinities were statistically significant.

All IndVals of the composite indices at all taxonomic resolutions were statistically significant at P < 0.01 (Table 3.6). Many IndVals, derived from composite indices, were greater than those of their individual component taxa. In particular, IndVals of ant species habitat ‘specialists’ increased considerably; some of them reaching 100%.

Difference between sampling methods Both pitfall trapping and litter extraction had significant discriminatory power (Table 3.1) and none of the habitat indicators showed contrary outcomes (rainforest versus pasture indicators) across the two methods (Tables 3.2-3.5). However, different methods often detected different indicator taxa and, when finer taxonomic resolution was employed, the differences became more evident. This suggests that there was complementarity between the two sampling methods at finer levels of taxonomic resolution. Indeed, the greatest increase in the ANOSIM Global R values was observed when pitfall traps and litter extraction were combined in the ant species dataset (Table 3.1). The only exception, where no complementarity was observed, was ant functional groups.

Additional results are also presented in Appendix 4a, b.

46

Table 3.4 Mean frequency scores and Indicator values (IndVals) of ant species habitat indicators. Only statistically significant indicators for each sampling method are shown. Pitfall traps Litter extraction RM1 PM1 IndVal2 Type3 RM1 PM1 IndVal2 Type3 Rainforest indicators Rainforest indicators Pheidole QM2 (ampla grp.) 1.75 0 91.7** S Pheidole QM2 (ampla grp.) 1.08 0 75.0** S Pheidole sp.2 (variabilis grp.) 1.75 0 75.0** S Monomorium tambourinense 1.08 0 66.7** S Anonychomyrma QM3 0.75 0 50.0* S Mayriella abstinens complex 0.75 0 50.0* S Notostigma foreli 0.67 0 50.0* S Lordomyrma QM1 0.50 0 41.7 S Pheidole QM1 (ampla grp.) 1.00 0 50.0* S Discothyrea QM1 0.58 0 33.3 S Monomorium tambourinense 0.67 0 41.7 S Hypoponera QM5 0.33 0 33.3 S Leptomyrmex erythrocephalus rufithorax 0.58 0 41.7 S Carebara QM2 0.50 0 33.3 S Pachycondyla QM2 (porcata grp.) 0.33 0 33.3 S Strumigenys harpyia 0.42 0 33.3 S Prolasius QM2 (bruneus grp.) 0.50 0 33.3 S Hypoponera sp.1 2.42 0.25 90.6** I Notoncus capitatus 1.00 0.08 53.8* I Rhytidoponera chalybaea 1.25 0.50 53.6 I

Pasture indicators Pasture indicators Rhytidoponera metallica 0 2.17 83.3** S Paratrechina QM2 (vaga grp.) 0 1.50 75.0** S Paratrechina QM2 (vaga grp.) 0 1.83 66.7* S Pheidole QM3 (grp. C) 0 1.08 66.7** S Pheidole QM3 (grp. C) 0 1.50 66.7* S Carebara QM1 0.25 0.92 45.8 I Cardiocondyla nuda 0 0.58 50.0* S Polyrhachis angusta 0 0.33 33.3 S Carebara QM1 0 0.83 50.0* I 1RM, PM, mean frequency scores in rainforest and pasture respectively. 2Indicator values (IndVals) and the results of their permutation test: * = P < 0.01, ** = P < 0.001, otherwise P < 0.05. 3S, I, habitat ‘specialists’ and ‘increasers’ respectively (see the Methods for more detail).

47

Table 3.5 Mean relative abundances and Indicator values (IndVals) of ant functional group habitat indicators. Only statistically significant indicators for each sampling method are shown. Pitfall traps Litter extraction RM1 PM1 IndVal2 Type3 RM1 PM1 IndVal2 Type3 Rainforest indicators Rainforest indicators Specialist 5.9 0.8 80.9** I Tropical climate 8.3 0 66.7** I predators specialists Generalised 37.9 9.0 80.8** I Cold climate 12.1 0.8 62.5* I Myrmicinae specialists Tropical climate 4.0 0.2 78.7** I Specialist 3.1 0 41.7 I specialists predators Cold climate 12.0 2.9 74.0* I specialists Dominant 9.5 2.0 48.0 I Dolichoderinae Pasture indicators Pasture indicators Opportunists 22.7 71.1 90.6** I Opportunists 3.5 34.4 75.8** I 1RM, PM, mean relative abundances in rainforest and pasture respectively. 2Indicator values (IndVals) and the results of their permutation test: * = P < 0.01, ** = P < 0.001, otherwise P < 0.05. 3S, I, habitat ‘specialists’ and ‘increasers’ respectively (see the Methods for more detail).

3.4 Discussion

Use of different taxonomic resolutions Datasets based on both coarse and fine levels of taxonomic resolution clearly separated rainforest from pasture in the ordinations. Ant species best segregated rainforest from pasture. However, the second best segregation was found using coarsely sorted arthropods, implying that an increasing level of taxonomic resolution does not necessarily show an increasing sensitivity of assemblage response in a simple manner. This result must be treated carefully, as alternative focal taxa (e.g. carabid-beetles, which are also popular bio-indicators; see Rainio & Niemela 2003) may show different sensitivities, resulting in different conclusions.

48 Table 3.6 Mean index levels and Indicator values (IndVals) of composite indices under each habitat indicator group. Composite indices of ‘specialists’ only (S), ‘increasers’ only (I) and combined ‘specialists’ and ‘increasers’ (S + I) are shown separately. IndVal was not calculated for a group containing less than two component taxa (T1 < 2). Pitfall traps Litter extraction T1 RM2 PM2 IndVal3 T1 RM2 PM2 IndVal3 Rainforest indices Rainforest indices Coarse arthropods (S) 2 0.69 0 83.3** Coarse arthropods (S) 1 - - - Coarse arthropods (I) 9 4.71 1.61 74.6** Coarse arthropods (I) 14 7.83 2.34 77.0** Coarse arthropods (S + I) 11 5.40 1.61 77.1** Coarse arthropods (S + I) 15 8.17 2.34 77.7**

Ant genera (S) 4 3.00 0 91.7** Ant genera (S) 1 - - - Ant genera (I) 2 3.92 1.92 67.1** Ant genera (I) 4 5.17 0.42 92.5** Ant genera (S + I) 6 6.91 1.92 78.3** Ant genera (S + I) 5 5.67 0.42 93.2**

Ant species (S) 9 8.00 0 100** Ant species (S) 8 5.25 0 100** Ant species (I) 2 2.25 0.58 72.8* Ant species (I) 1 - - - Ant species (S + I) 11 10.25 0.58 94.6** Ant species (S + I) 9 7.67 0.25 96.8** Pasture indices Pasture indices Coarse arthropods (S) 0 - - - Coarse arthropods (S) 0 - - - Coarse arthropods (I) 3 0.79 2.04 72.0** Coarse arthropods (I) 1 - - - Coarse arthropods (S + I) 3 0.79 2.04 72.0** Coarse arthropods (S + I) 1 - - -

Ant genera (S) 2 0 0.92 58.3* Ant genera (S) 0 - - - Ant genera (I) 2 0.67 2.67 66.7* Ant genera (I) 1 - - - Ant genera (S + I) 4 0.67 3.58 70.3* Ant genera (S + I) 1 - - -

Ant species (S) 5 0 6.42 91.7** Ant species (S) 2 0 2.58 91.7** Ant species (I) 1 - - - Ant species (I) 1 - - - Ant species (S + I) 6 0 7.25 91.7** Ant species (S + I) 3 0.25 3.50 93.3** 1Number of taxa used for that composite indicator. 2

49 RM, PM, mean index levels in rainforest and pasture respectively. 3Indicator values (IndVals) and the results of their permutation test: * = P < 0.01, ** = P < 0.001.

As far as ants are concerned, my results suggest that if sorting and identification to species level is not possible due to limited taxonomic knowledge and/or resources, it may be more rewarding to investigate the whole or a subset of the arthropod assemblage at a coarse taxonomic resolution rather than sorting ants to an intermediate resolution such as genus level. My results broadly concur with those of Nakamura et al. (2003), who investigated the utility of soil and litter dwelling arthropods to assess the state of rainforest restoration. They found that, compared with assemblages based on ant genera, a subset of coarse arthropods better represented the state of habitat recovery from disturbed (pasture) to developing rainforest.

The use of coarse taxonomic resolutions to assess invertebrate responses to habitat conditions has been advocated by a number of authors (see Pik et al. 1999), for the rapid assessment of environmental health, especially where taxonomic knowledge is lacking. This approach has been used to monitor environmental changes in Japanese temperate forests and Kalimantan rainforests (Aoki & Harada 1982; Sakaino et al. 2002). However, unless parallel studies at the species level are carried out, the results should be treated with caution. For example, when a habitat ‘increaser’ is found at a taxonomic resolution higher than the species level, it is possible that this taxon is composed of individual species which may respond differently to habitat conditions. My study found that Pheidole was a rainforest ‘increaser’ at genus level; however, Pheidole QM1 and QM3 were detected as rainforest and pasture ‘specialists’ respectively.

Use of composite indices One of my aims is to generate a useful indicator that can measure the habitat condition of vegetation stages that are intermediate between rainforest and pasture. One of the biggest constraints in the search for reliable habitat indicators is the patchy distributions of target organisms (Giller 1996; Veech et al. 2003). Ants are no exception to this. Other studies have found the distribution of ant species to be quite patchy and their assemblages varied over scales ranging from a few kilometres down to 1 m2 within a homogenous habitat (Wilson 1958; Kaspari 1996; Soares & Schoereder 2001). Nakamura et al. (2003) also found patchy distributions of ant genera in all habitat types studied (five rainforest remnant, five pasture and 10 revegetated sites located across a catchment area of over 150 km), resulting in no distinct differences in the composition of ant assemblages among these habitat types. The present study was carried out over a much smaller geographical

50 scale with a large number of sampling sites; however, many habitat indicators still showed patchy distributions. This was particularly prominent for ant species. Many ant species were habitat ‘specialists’ (high specificity), but were encountered only sporadically (low fidelity), rendering individual indicator species unreliable.

One way to overcome this problem is to generate composite indices (cf. Kitching 1993a) by combining information from sets of species that are characteristic of the habitat in question. In contrast to individual ant indicator species, IndVals generated using composite indices were well over the benchmark of 70%, with greatly increased fidelity as at least one species belonging to the composite index occurred at almost all sites within their associated habitat.

One may argue, however, that the poor reliability of individual indicator taxa was simply due to insufficient sampling intensity at each site. More intense sampling may increase the chances of catching rarer indicators, hence increasing the fidelity of individual indicator species. In reality, monitoring programs are often under-funded and cannot afford to conduct intense sampling at a site. Further, if the use of composite habitat indices can overcome the problem of patchy species distribution, it may be more rewarding to cover larger number of sites rather than monitoring a few intensely.

The use of composite indices not only increases the IndVals, but also reduces the risk of relying on one or a few individual taxa, whose value as bio-indicators may be compromised due to seasonality or the geographically aggregated nature of species distributions. This approach has another great advantage if the aims of habitat assessment are to monitor changes in assemblages of species that are restricted to particular habitat types that are to be conserved or restored. A much stronger indicator signal can be generated by excluding species that contribute ‘noise’ to the data, such as generalist species. Furthermore, composite indices are more economical in terms of efficiency of communication, since they involve fewer components, compared to the use of numerous IndVals for individual taxa. Such efficiency is important to managers and decision-makers who wish to rapidly assess the progress of ecological degradation or restoration.

51

52 4 EFFECTS OF ISOLATION AND INOCULATION

4.1 Introduction

Forest clearance is one of the most pervasive of human-induced disturbances which threatens terrestrial biodiversity (Sala et al. 2000). Consequently, a considerable amount of recent effort has been put into restoring the forest cover of previously cleared areas (Young et al. 2005; Catterall et al. 2007). One of the important criteria for judging such restoration as successful from a conservation viewpoint, is that the revegetated area should have a characteristic assemblage of species, comparable with that occurring in undisturbed reference sites (White & Walker 1997; Catterall et al. 2004; SERI 2004). Reforestation activities are often conducted with the underlying assumption that fauna will recolonise from surrounding areas naturally, as successional development of re-planted vegetation proceeds (Majer 1990). However, studies of the factors required for successful recolonisation by animals have been neglected until recently (Scott et al. 2001).

Tree planting to restore areas of forest often creates ‘islands’ of habitat suited to forest-dependent organisms, surrounded by a hostile ‘sea’ of cleared land (Hobbs 1993). The rate at which these habitat patches are colonised by species that depend on forest may be strongly affected by their degree of isolation from ‘source’ forest habitat (cf. MacArthur & Wilson 1967). In addition, the nature of the matrix within which the patches are established may also be important: arable land or pasture surrounding restored forest patches, for example, may allow some dispersing taxa to persist, despite being unsuitable as habitat for many forest-dependent species (As 1993; Driscoll 2005). The process may be further complicated by the fact that the surrounding matrix may accommodate non-forest taxa that can colonise and use the newly restored sites rapidly, out-competing any subsequent colonisers from existing forest habitats.

Colonisation processes are also influenced by the autecology of the various target taxa, especially their different dispersal abilities (den Boer 1990). Mobile arthropods, including large-winged, wind-dispersed, and phoretic species, for example, may reach isolated habitats relatively readily (see den Boer 1990; Athias-Binche 1993; Compton 2002). In contrast, wingless or short-winged arthropods, well-represented in old-growth forests (den Boer 1990), must walk through the surrounding matrix to reach the restored

53 sites. These less mobile species must be able to withstand the potentially hostile conditions of the matrix to ensure successful colonisation. If they are incapable of this, or simply avoid entering the matrix, then the natural recolonisation of isolated restoration sites becomes improbable.

To overcome the limited dispersal abilities of less mobile animals, inoculation (translocation of animals from undisturbed habitats to restored areas) may be useful (Bradshaw 1996; Brady et al. 2002). A number of studies have reported increased species diversity and abundance within macro-invertebrate assemblages, including less mobile taxa, following inoculation in restored aquatic habitats (e.g. Brown et al. 1997; Brady et al. 2002). In terrestrial systems, inoculation has also been suggested as a restoration tool (e.g. Recher 1989; Keesing & Wratten 1998; Bowie et al. 2003). In practice, this has been largely restricted to the re-introduction of charismatic taxa (Thomas 1991), or of a limited number of species which are considered to play important roles in ecosystem functioning (e.g. earthworms for nutrient cycling, Curry & Good 1992). The method has rarely been used in the attempted re-establishment of whole faunal assemblages (but see Mortimer et al. 2002).

This chapter investigates the effect of isolation on the patterns of colonisation of restored Australian rainforest in a matrix of cleared pasture by assemblages of soil and litter arthropods. Rapid restoration of rainforest vegetation is often carried out in Australia by densely planting the seedlings of rainforest trees into patches of former arable land or pasture (Kanowski et al. 2003; Catterall et al. 2007). To investigate recolonisation rates under controlled conditions, I used a field experiment, in which the environmental characteristics (relevant to soil and litter arthropods) of restored rainforest patches were simulated within plots of shaded litter at varying distances from rainforest edges. The experiment included inoculation of the most isolated plots.

The experiment represents a reductionist approach; however, the great diversity and levels of uncontrolled variation inherent in pre-existing restoration plots (the usual subjects of such studies) demand parallel simple experimental manipulations. I test the hypothesis that the abundance and diversity of rainforest-associated arthropods in restored patches will decrease with distance from rainforest remnants. Further, plots which receive inoculations of forest litter are expected to show increased abundance and diversity of rainforest taxa.

54 4.2 Methods

Study area The study was undertaken on the Maleny plateau, in the Sunshine Coast hinterland of eastern Australia (26° 40’- 50’ S, 152° 45’- 53’ E, elevation 350 to 530 m). At the time of the study, large parts of the Maleny plateau were pasture used for grazing dairy or beef cattle. Mean daily maximum and minimum temperatures in mid summer (January) are 28.9 and 18.8 ° C respectively, and 19.5 and 7.1 ° C in mid winter (July). Frost is common in winter (June-August). Average annual rainfall in the region is 1851 mm, with most (170-310 mm per month) falling between December and April (climate data averaged over at least 50 years up to 2004, obtained from the Australian Bureau of Meteorology).

Five experimental sites were dispersed across a study region of approximately 170 km2; all within 13 km of the township of Maleny (Figure 2.1). The minimum distance between sites was 1 km; most sites were > 2 km apart. Each site was comprised of pasture abutting rainforest remnants. The pasture at four of the sites was heavily grazed by cattle and/or horses, and the other was lightly grazed (Appendix 2). Adjacent rainforest remnants were either old regrowth (age ca.100 years) or remnants that had been selectively logged until recently. Edges of the rainforest remnants were fenced to exclude livestock. Further details of study sites are provided in Section 2.1.

Experimental design At each site, a series of experimental plots was established to simulate the environmental conditions experienced by soil and litter arthropods within areas of replanted rainforest. Each plot consisted of a 3 x 3 m quadrat. At each site, experimental plots were placed at 0, 15, 100 and ca. 400 m from the rainforest edge (Figure 2.2). The plot at the edge of the rainforest (i.e. 0 m) was located in pasture at approximately two metres from the edge of the rainforest undergrowth or canopy cover, whichever was closer. An additional plot was established at 400 m from the rainforest edge in order to test the efficacy of inoculation. A control plot was established within each rainforest remnant, at least 30 m from the rainforest edge.

Construction of experimental plots simulated actual replanting procedures (Kanowski et al. 2003; Big Scrub Rainforest Landcare Group 2005; Catterall et al. 2007), based on discussions with revegetation practitioners in the study region. During rainforest

55 replanting in the Maleny region, pasture is first sprayed with broad spectrum herbicide. After the grass has died a dense layer of mulch (typically of hay or woodchip) is added, and seedlings of rainforest trees are planted around 1-2 m apart (Kanowski et al. 2003; Catterall et al. 2004; Catterall et al. 2007). In the Maleny region, a canopy cover of 80 % develops within 10 years as the tree seedlings grow rapidly (Kanowski et al. 2003; Nakamura et al. 2003).

Construction of experimental plots began with the application of herbicide (Roundup® Biactive™, 7.2 g/L glyphosate) over each 3 x 3 m plot (Figure 2.3). It should be noted that an additional test found no short- or long-term impacts of the herbicide on soil and litter arthropods inhabiting rainforest litter (Chapter 7). Three weeks after the application, all visible vegetation (dead and alive) was removed by hand. Steel posts, at a height of 1.2 m, were then erected at the corners, and fenced with barbed wire to exclude livestock. To simulate shading by the developing tree canopy, Sarlon® shadecloth (rated at 90 % protection from insolation) was placed over the top of the quadrat and 20-40 cm down each side. The shadecloth was cut in 15-20 places with a slit length of 10-15 cm to permit sunflecks and throughfall of rain.

To provide a potential habitat for soil and litter arthropods, a mulch of woodchip and leaf material, 10-15 cm in depth, was placed in a square area (2.5 x 2.5 m) established inside each quadrat. The area was bordered by wire netting (40 cm high, hexagonal mesh size of 1.5 cm [maximum height] x 2 cm [maximum width]) to minimise loss of mulch due to wind or disturbance by wildlife. The mulch material was derived from lopping vegetation around powerlines in various locations within about 150 km of the study region, conducted by a local contractor of the electricity provider. Woodchip mulch comprised a mix of foliage and wood derived from rainforest and eucalypt species. Before mulch was added to the experimental plots, it was steam-treated for 100 minutes to kill existing arthropods. The efficacy of the treatment was tested by running approximately 10 litres of freshly steamed mulch in Tullgren funnels. The resulting extracts were checked every 4-8 hours for a total of 5 days and no live arthropods were found. Sterilised mulch was distributed to the experimental plots within seven days after steam-treatment. Before its distribution, mulch was stored beneath plastic sheets to minimise any casual arthropod colonisation.

56 Control plots inside rainforest remnants were constructed in exactly the same manner as other plots except that a tent of plastic bird netting (mesh size of 2 x 2 cm) covered the entire plot, minimising the natural fall of rainforest leaves and twigs into the experimental area.

The following protocol was used for the inoculation treatments. Paired plots, 80-100 m apart, were established at approximately 400 m away from the rainforest edge in each site. Both plots were identical in terms of construction and mulch, except that one of them received inoculum from the rainforest interiors. Topsoil and leaf litter were collected from within the rainforest remnant at each of the five sites and mixed thoroughly to produce the required inoculum. Approximately 20 litres of inoculum (about 2% of the total mulch used at a plot) was buried in a surround of steam-treated woodchip mulch at the centre of the ‘inoculum’ plot. In order to minimise disturbance, the inoculum was handled carefully (i.e. avoiding compression and temperature extremes), and the entire task was carried out within 12 hours. Six litres of the inoculum was extracted using a Tullgren funnel to determine the composition of the added arthropod assemblage.

The entire construction of the field experiment took approximately three months and was completed in August 2003.

Sampling methodology Sampling was carried out between 9 April and 7 May 2004 (approximately nine months after the plots were established). At each site, arthropods were collected using two methods: pitfall trapping and litter extraction. Pitfall trapping samples the relative density of active epigaeic arthropods, whereas litter extraction samples the relative abundance of sedentary arthropods that live in litter and topsoil (Majer 1997). Four 4.4 cm diameter pitfall traps (120 ml) were installed approximately 80 cm from the centre of the plot at even intervals (Figure 2.3). Vials were partly filled with 70% ethanol with a small amount of glycerol, and operated for five days. Before data analysis, samples from the four pitfall traps were pooled. Litter extraction was carried out by collecting 1 litre of litter and surface soil (to a depth of 1 to 2 cm) in small amounts evenly over the entire plot area (approx. 20% surface soil and 80% litter by volume). Samples were placed in Tullgren funnels within 12 hours of sampling, and extracted for 4.5 days using 40 W clear light bulbs.

57 During sampling, care was taken to avoid cross-contamination among the experimental plots. All footwear was covered with thick polythene film, and researchers were thoroughly brush-cleaned before and after each plot was visited. On each sampling day, the furthest plots (400 m) were sampled first. Plots in pasture were not visited once control plots within the rainforest remnants had been sampled.

Identification of arthropods was to Order except a) Hymenoptera which were split into Formicidae and others and b) myriapods to Class. Acari, Collembola and Diptera (with the exception of litter-extracted dipterans) were not sorted due to their high abundance and ubiquitous occurrence regardless of habitat type (rainforest or pasture). Ants (Hymenoptera: Formicidae) were selected as a target group, and sorted to species. With assistance from Alan Andersen (CSIRO Sustainable Ecosystems), ant species were assigned to functional groups on the basis of their presumed responses to environmental stress and disturbance (Andersen 2000a; Brown 2000).

During the arthropod sampling, the ground temperature was recorded at 30 minute intervals for 5.5 days (14 to 20 April 2004) using temperature loggers (HOBO® Temperature Data Logger, Onset Computer Corporation, MA), deployed at the experimental plots and pasture and rainforest reference habitats in one site. The temperature sensor was placed on the ground surface, beneath the litter, in the centre of each plot. To measure soil moisture content, topsoil (up to 2-3 cm in depth) of approximately 50 cm3 was hand collected once from all plots at all five sites. Small amounts of topsoil were collected evenly over the plot area, or 2.5 x 2.5 m quadrats established within rainforest and pasture of each site. The collection dates were between 15 April and 7 May 2004. The soil was kept in an air tight plastic bag, and then weighed before and after it was oven dried for 24 hours at 105 C˚.

Data analysis Three different datasets were analysed: arthropods sorted to Orders/Class (referred to as ‘coarse arthropods’ hereafter), ant species and ant functional groups. Each dataset was further divided into two separate subsets comprising data sampled using pitfall traps and litter extraction. Abundances of coarse arthropods were log-transformed before analysis. Prior to analyses, abundances of ant species were scored on a seven-point scale following Andersen et al. (2003): 1 = 1, 2 = 2-5, 3 = 6-20, 4 = 21-50, 5 = 51-100, 6 = 101-1000, 7 = >1000 individuals. Abundances within ant functional groups were expressed as

58 proportions of all ants (based on raw abundance) at each plot, and arcsine-transformed before analysis.

In addition to the data obtained from the present study, an earlier survey (Chapter 3) provided baseline information on the arthropod assemblages in pasture and rainforest habitats of the study sites. This survey was carried out in the same region and in a similar season (8 January to 6 May 2003). Arthropods were collected from three sampling points (positioned 15-30 m apart from each other) at each of 12 sites across Maleny (N = 36), including the five sites used for the present study. Although arthropod sampling was carried out over a larger area than that of the present study (each sampling point comprising a circular area of 3 m radius), sampling and sorting protocols were otherwise identical to those of the present study, so that direct comparison between the results of the baseline survey and the present study is possible.

This survey identified taxa that were characteristic of either rainforest or pasture habitats. These habitat indicator taxa were classed as either habitat ‘specialists’ (those which were found exclusively in their preferred habitat) or habitat ‘increasers’ (those which were found in both habitat types but were significantly more abundant in one, see Chapter 3 for more details). However, many individual habitat indicator taxa were of limited usefulness due to their patchy distributions. To develop a more robust indicator statistic, additional ‘composite rainforest/pasture indices’ (Chapter 3) were generated for the selected datasets (viz. coarse arthropods, ant species). To obtain the composite indices, the abundance values were summed across all taxa indicative of either rainforest (to calculate composite rainforest indices) or pasture (to calculate composite pasture indices) at each site, providing a single value which quantifies the extent to which a site is rainforest-like or pasture-like, in terms of its arthropod assemblage. Composite indices were calculated separately for coarse arthropods and ant species collected by either pitfall traps or litter extraction. Before calculating composite indices of coarse arthropods, abundance values of each of the component taxa were first individually range-standardised (site-specific abundance minus minimum abundance across all sites / maximum minus minimum) to remove the effects of large differences in taxon-specific abundance. This gave values between 0 and 1 for each taxon at each site. This range-standardisation procedure was not carried out for ant species, as most indicator species had a similar range in abundance score values (most ranging 0-3, with a maximum score of 4).

59 Repeated-measures ANOVA were performed to test for differences in a number of variables (taxon/species richness, levels of composite pasture/rainforest indices, abundances of individual arthropod taxa, ant species and functional groups, and soil moisture content) among different distances (-30 m [control], 0 m, 15 m, 100 m, 400 m) from the rainforest edge. Distance was the within-subject factor in the repeated-measures design, and sites were the subjects. Each variable was analysed using two separate repeated-measures ANOVAs: (1) a test for differences across habitat boundaries (rainforest [-30 m], edge habitat [0 m], pasture [15 m]); (2) a test for differences among distances away from the rainforest edge (15 m, 100 m, 400 m). Degrees of freedom for all F tests were corrected by Huynh-Feldt Epsilon value, whether or not each variable tested violated the assumption of sphericity, as recommended by Quinn and Keough (2002). Paired t-tests were used to test for differences between inoculation and no-inoculation (400 m) plots. To permit direct comparison with pasture and rainforest reference sites, I also present arthropod data from the baseline survey obtained at the same five sites (one of the three sampling points randomly selected at a site).

Multivariate analyses were carried out to investigate differences in the arthropod assemblages among the plots with different experimental treatments. Non-metric multi-scaling ordination (NMDS) was used with PRIMER v.5 software (Clarke 1993). All NMDS ordinations were performed on Bray-Curtis similarity matrices calculated between each site pair, with 10 random restarts. Analysis of Similarity (ANOSIM: Clarke 1993), with 999 permutations, was used to test for significance of discriminatory power between different treatments.

Although multiple statistical testing was involved in this study, family-wise Type I error was not controlled and the significant P threshold was set at 0.05. This is because avoidance of Type II errors was important for the purpose of this study, to screen for taxa which responded to the treatment effects (Roback & Askins 2005; see also Moran 2003). The risk of Type II error was high, because of the small sample size (5 replicate sites).

4.3 Results

A total of 5363 arthropods were sampled (3888 from pitfall traps and 1475 from litter extraction). Twenty-three coarse arthropod taxa were identified. Forty-nine ant species were identified, but their occurrence was patchy (see also Appendix 3b). This was

60 particularly so for litter extraction – seven plots were excluded from the following analyses, as no ants were extracted from six plots and in one 400 m plot only one ant species (Pyramica membranifera (Emery)), which was unique to this plot, was sampled. At one of the 400 m plots, over 60% of the total ant captures from pitfall traps comprised invasive coastal brown ants (Pheidole megacephala (Fabricius)).

Among nine ant functional groups recorded, Cryptic species, Opportunists and Generalised Myrmicinae were very abundant, comprising over 95% of total ant catches. Dominant Dolichoderinae (numerically dominant species in hot and open environments, see Andersen 2000a) was almost absent from the experimental plots. One functional group (pitfall-trapped Generalised Myrmicinae) responded significantly in relation to distance (across habitat boundary, P = 0.010) and the inoculation treatments (P = 0.027); however, the patterns were not clearly interpretable. None of the other results from either multivariate or univariate analyses of assemblages of ant functional groups were statistically significant.

4.3.1 Isolation

Arthropod assemblage composition NMDS ordinations (Figure 4.1a, b) and ANOSIM analyses (Table 4.1) of differences in the assemblage composition of coarse arthropods showed that baseline pasture and rainforest sites were clearly separated in both pitfall traps and litter extraction. For pitfall-trapped coarse arthropods (Figure 4.1a), control (-30 m) and edge plots (0 m) were within the rainforest cluster (see also Table 4.1). All plots located away from the rainforest edge (15 m, 100 m, 400 m) were separated from the rainforest cluster (Figure 4.1a), and there were no significant differences among them (Table 4.1). Litter extraction (Figure 4.1b) showed similar patterns to those found from the pitfall traps except that control (-30 m) and edge plots (0 m) were not within the rainforest cluster (Table 4.1).

The ant species composition was also significantly different between pasture and rainforest in both pitfall traps and litter extraction (Figure 4.1c, d, Table 4.2). For pitfall-trapped ant species (Figure 4.1c), the control plots (-30 m) were within the rainforest cluster (see also Table 4.2). Edge plots (0 m) were slightly off the rainforest cluster (Figure 4.1c), but did not differ significantly from rainforest (Table 4.2). More distant plots (15 m, 100 m, 400 m) were more similar to pasture than rainforest (Figure 4.1c, Table 4.2). Litter-extracted ants showed broadly similar, but less clear, patterns 61 (Figure 4.1d, Table 4.2). This result may have been due to the lower numbers and patchier distribution of ants sampled using litter extraction.

a Coarse arthropods (Pitfall traps) b Coarse arthropods (Litter extraction)

c Ant species (Pitfall traps) d Ant species (Litter extraction)

Legend

Rainforest (RF) reference Pasture (P) reference Control (-30 m) Edge (0 m) 15 m 100 m 400 m

Figure 4.1 NMDS ordination of experimental plots at various distances from the rainforest edge, and rainforest and pasture reference habitats. Assemblages sampled using different sampling methods (pitfall traps, litter extraction) are shown separately.

62 Table 4.1 ANOSIM global R and pair-wise R values comparing coarse arthropod assemblage compositions between treatments, for each trapping method. Significant P values (< 0.05) are shown in bold. Pitfall traps Litter extraction (Global R: 0.43, (Global R: 0.68,

P: < 0.001) P: < 0.001) R R Treatment pairs statistic P statistic P Rainforest v Pasture 0.55 <0.001 0.67 <0.001

-30 m v Rainforest 0.05 0.348 0.39 0.005 0 m v Rainforest 0.05 0.343 0.75 <0.001 15 m v Rainforest 0.51 0.002 0.98 <0.001 100 m v Rainforest 0.49 0.005 0.96 <0.001 400 m v Rainforest 0.38 0.018 0.97 <0.001

-30 m v Pasture 0.56 0.003 0.70 <0.001 0 m v Pasture 0.35 0.018 0.71 <0.001 15 m v Pasture 0.44 0.011 0.81 <0.001 100 m v Pasture 0.38 0.016 0.78 <0.001 400 m v Pasture 0.40 0.006 0.83 <0.001

0 m v -30 m 0.60 0.008 0.12 0.167 15 m v -30 m 0.53 0.008 0.42 0.032 100 m v -30 m 0.58 0.008 0.44 0.016 400 m v -30 m 0.59 0.008 0.60 0.016

15 m v 0 m 0.20 0.071 0.09 0.175 100 m v 0 m 0.30 0.032 0.09 0.214 400 m v 0 m 0.26 0.087 0.02 0.452

100 m v 15 m -0.09 0.722 0.05 0.286 400 m v 15 m 0.00 0.460 0.05 0.286 400 m v 100 m -0.02 0.508 0.03 0.373

Taxon richness of coarse arthropods from both pitfall traps and litter extraction decreased substantially from edge (0 m) to 15 m plots (Figure 4.2a, Table 4.3a), but showed little change with increasing distance into pasture (15 m, 100 m, 400 m; see Table 4.3a). Similar patterns were found for ant species richness; although no results from litter extraction were statistically significant (Figure 4.2b, Table 4.3b).

The composite rainforest index based on coarse arthropods showed a similar pattern to that of taxon/species richness – a considerable decrease in the value as soon as plots were located away from the rainforest edge (Figure 4.2c, Table 4.3a). For ant species (Figure

63 4.2d, Table 4.3b), levels of the composite rainforest index declined at the rainforest edge and became zero at almost all plots within pasture.

Table 4.2 ANOSIM global R and pair-wise R values comparing ant species assemblage compositions between treatments, for each trapping method. Significant P values (< 0.05) are shown in bold. Pitfall traps Litter extraction (Global R: 0.44, (Global R: 0.39,

P: < 0.001) P: < 0.001) R R Treatment pairs statistic P statistic P Rainforest v Pasture 0.63 <0.001 0.54 <0.001

-30 m v Rainforest -0.01 0.500 0.12 0.200 0 m v Rainforest 0.24 0.064 -0.02 0.517 15 m v Rainforest 0.65 <0.001 0.62 <0.001 100 m v Rainforest 0.47 0.002 0.63 0.002 400 m v Rainforest 0.51 <0.001 0.55 0.016

-30 m v Pasture 0.66 <0.001 0.40 <0.001 0 m v Pasture 0.39 0.004 0.13 0.119 15 m v Pasture 0.22 0.047 0.22 0.024 100 m v Pasture 0.13 0.113 0.12 0.146 400 m v Pasture 0.27 0.017 0.15 0.143

0 m v -30 m 0.47 0.016 -0.14 0.849 15 m v -30 m 0.45 0.016 0.12 0.206 100 m v -30 m 0.62 0.008 0.30 0.125 400 m v -30 m 0.62 0.008 0.10 0.429

15 m v 0 m 0.23 0.024 0.07 0.343 100 m v 0 m 0.10 0.175 0.42 0.057 400 m v 0 m 0.13 0.143 0.25 0.267

100 m v 15 m -0.03 0.571 -0.17 0.829 400 m v 15 m -0.02 0.524 -0.13 0.733 400 m v 100 m -0.19 0.952 0.21 0.300

The composite pasture index of pitfall-trapped coarse arthropods showed that the index levels of experimental plots at any distances were suppressed to similar levels to those found in the rainforest samples (Figure 4.2e, Table 4.3a). Although not statistically significant, pitfall-trapped ant species had increased levels of the composite pasture index in plots located at 15 m or more from the rainforest edge (Figure 4.2f, Table 4.3b).

64

a Coarse arthropod taxon richness b Ant species richness 16 8 12 6

8 4

4 Taxon richness 2 Species richness Species

0 0 RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P

c Coarse arthropod composite rainforest index d Ant species composite rainforest index 6 10 5 8 4 6 3 4 Index level level Index 2 level Index

1 2

0 0 RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P

e Coarse arthropod composite pasture index f Ant species composite pasture index

1.6 6

1.2 4 0.8 Index level Index Index level level Index 2 0.4

0 0 RF -30 -30m m 0m0 m 15 15m m 100 100m m 400m400 m P RF -30 -30m m 0m0 m 15 15m m 100 100m m 400m400 m P Distance from rainforest edge Distance from rainforest edge

Figure 4.2 Mean taxon richness and composite rainforest and pasture indices of coarse arthropods and ant species across the five different distances (-30 m, 0 m, 15 m, 100 m, 400 m) from rainforest edge. Values for rainforest (RF) and pasture (P) reference habitats (sampled the previous year) are also shown. Open bars, pitfall traps; closed bars, litter extraction.

65

66 Table 4.3 Summary results of repeated measures ANOVA for (a) coarse arthropods and (b) ant species. The results of individual habitat indicators are shown only when their abundance (or abundance scale) responded significantly to the experimental treatments. Across the habitat boundary, tests compare plots at -30 m, 0 m, and 15 m. Away from habitat boundary, tests compare plots at 15 m, 100 m, and 400 m. Significant P values (< 0.05) are shown in bold. Pitfall traps Litter extraction Across habitat Away from habitat Across habitat Away from habitat

boundary boundary boundary boundary F P F P F P F P a. Coarse arthropods Taxon richness 9.02 0.016 1.38 0.306 6.61 0.034 0.85 0.461 Composite rainforest index 29.65 <0.001 2.42 0.151 3.53 0.125 0.11 0.839 Composite pasture index 0.37 0.701 0.45 0.654 NA† NA†

Coleoptera (Rainforest ‘increaser’) 4.68 0.048 0.73 0.511 2.91 0.113 2.91 0.112 Heteroptera (Rainforest ‘increaser’) 18.04 0.002 0.23 0.798 8.99 0.009 NA† Pauropoda (Rainforest ‘increaser’) 39.08 0.001 NA† 12.76 0.010 1.46 0.289 Pseudoscorpionida (Rainforest ‘increaser’) 6.46 0.021 NA† NA† NA† Diplopoda (Rainforest ‘increaser’) 0.76 0.498 NA† 8.76 0.010 NA† Symphyla (Rainforest ‘increaser’) NA† NA† 7.03 0.030 NA† Dermaptera (Rainforest ‘increaser’) 8.80 0.016 3.16 0.097 1.16 0.361 2.05 0.222 Isopoda (Rainforest ‘increaser’) 10.34 0.022 2.67 0.129 0.47 0.640 0.19 0.834

b. Ant species Species richness 10.66 0.006 0.47 0.631 2.39 0.154 0.23 0.693 Composite rainforest index 14.47 0.002 NA† 4.19 0.076 NA† Composite pasture index 3.10 0.101 0.41 0.678 0.74 0.508 1.36 0.312

Pheidole QM2 (Rainforest ‘specialist’) 14.24 0.020 NA† NA† NA† Rhytidoponera metallica (Pasture ‘specialist’) 7.30 0.016 1.30 0.324 NA† NA† † Tests were not carried out for some variables that occurred at less than four experimental plots within each test.

Habitat indicators Among coarse arthropod taxa, there were 15 rainforest ‘increasers’, and three pasture ‘increasers’ (there were no rainforest or pasture ‘specialists’ in the samples). The abundances of six rainforest ‘increasers’, namely Coleoptera, Heteroptera, Pauropoda, Pseudoscorpionida, Diplopoda and Symphyla were significantly reduced at all locations outside of the rainforest edge (Figure 4.3a-f, Table 4.3a). Abundances of two rainforest ‘increasers’ (Dermaptera, Isopoda) varied significantly in a manner inconsistent with distance (Figure 4.3g, h); the remaining seven rainforest ‘increasers’, and all three pasture ‘increasers’ showed no significant responses to distance.

Among ant species, there were 14 rainforest ‘specialists’ and three rainforest ‘increasers’, and three pasture ‘specialists’ and one pasture ‘increasers’. However, due to the patchy occurrences of most of these ant species, statistical tests were possible for only four rainforest and three pasture indicators (occurring at > 4 plots within each test). Among the rainforest indicators, only Pheidole QM2 (rainforest ‘specialists’) responded significantly to distance (Figure 4.4a, Table 4.3b), and occurred only in control plots. Although not significant, one other rainforest ‘specialist’ and two ‘increasers’ (Pheidole QM1, Hypoponera sp.1, Rhytidoponera chalybaea Emery) occurred in control as well as edge plots, but did not occur at any of the more distant plots in pasture. In contrast, these distant plots were colonised by pasture ‘specialists’, including Rhytidoponera metallica (Smith) (Figure 4.4b, Table 4.3b) and Pheidole QM3 (not statistically significant). The edge plots (0 m) were also invaded by these pasture ‘specialists’ but to a much lesser degree. Responses of the other pasture indicators were not statistically significant.

67

a Coleoptera (Rainforest ‘increaser’) b Heteroptera (Rainforest ‘increaser’) 60 40 50

40 30 30 20 Abundance Abundance Abundance 20 10 10

0 0 RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P

c Pauropoda (Rainforest ‘increaser’) d Pseudoscorpionida (Rainforest ‘increaser’) 6 16 5

12 4 3 8 Abundance Abundance Abundance Abundance 2 4 1

0 0 RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P

e Diplopoda (Rainforest ‘increaser’) f Symphyla (Rainforest ‘increaser’) 16 8

12 6

8 4 Abundance Abundance Abundance 4 2

0 0 RFRF -30m-30 m 0m0 m 15 15m m 100 100m m 400 400m m P RF -30 -30m m 0 0m m 15 15m m 100 100m m 400 400m m P P

g Dermaptera (Rainforest ‘increaser’) h Isopoda (Rainforest ‘increaser’) 10 20 8 15 6 10 Abundance 4 Abundance 2 5

0 0 RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P Distance from rainforest edge Distance from rainforest edge

Figure 4.3 Abundances (mean, SE) of the individual arthropod taxa that showed statistically significant differences across the five different distances (-30 m, 0 m, 15 m, 100 m, 400 m) from the rainforest edge. Values for rainforest (RF) and pasture (P) reference habitats (sampled the previous year) are also shown. Open bars, pitfall traps; closed bars, litter extraction.

68

a Pheidole QM2 (Rainforest ‘specialist’) b Rhytidoponera metallica (Pasture ‘specialist’) 3 4

3 2

2 1

Abundance scale scale Abundance 1 Abundance scale scale Abundance

0 0 RF -30 -30m m 0m0 m 15 15m m 100 100m m 400 400m m P RFRF -30 -30m m 0 0m m 15 15m m 100 100m m 400 400m m P Distance from rainforest edge Distance from rainforest edge

Figure 4.4 Abundance scores (mean, SE) of the individual ant species that showed statistically significant differences across the five different distances (-30 m, 0 m, 15 m, 100 m, 400 m) from the rainforest edge. Values for rainforest (RF) and pasture (P) reference habitats (sampled the previous year) are also shown. Open bars, pitfall traps; closed bars, litter extraction.

Environmental variables Mean ground-level temperature within the rainforest reference habitat was slightly lower than in pasture (rainforest 18.3 ˚C, pasture 20.0 ˚C), at the single site where temperatures were measured. Experimental plots outside the rainforest all showed similar mean temperatures to pasture and to one another, irrespective of their position (0 m 20.0 ˚C, 15 m 20.8 ˚C, 100 m 19.7 ˚C, 400 m 20.0 ˚C); within the rainforest, the temperature of the control plot was intermediate (-30 m 19.2 ˚C). The differences were maintained consistently across the five days of temperature monitoring. Diurnal temperature fluctuations were much greater in pasture reference habitats than in rainforest (coefficient of variation: rainforest 3.50%, pasture 6.24%). Temperature fluctuations at experimental plots across different distances were similar to, or lower than in rainforest (-30 m 1.16%, 0 m 3.64%, 15 m 3.03%, 100 m 3.79%, 400 m 2.02%). The soil moisture content was higher in pasture than rainforest (Figure 4.5). Experimental plots located at the rainforest edge had conspicuously lower moisture contents than both rainforest and pasture, while other experimental plots had mean values intermediate between rainforest and pasture (ANOVA across all of the distances: F = 12.79, P < 0.001).

69

60 50 40 30 20 10

Soil moisture Soil moisture (%) content 0 RF -30 -30 m 0 m m 15 mm 100 100 mm 400 400 m m P Distance from rainforest edge

Figure 4.5 Soil moisture content (mean, SE) across the five different distances (-30 m, 0 m, 15 m, 100 m, 400 m) from rainforest edge, and for rainforest (RF) and pasture (P) reference habitats, across all sites. All samples were taken in 2004.

4.3.2 Inoculation

Extracts from six litres of raw inoculum contained a diverse array of rainforest taxa, with some ant queens (four individuals of Hypoponera sp.1, one each of Lordomyrma QM1 and Strumigenys harpyia Bolton) that might have been able to establish new colonies within the inoculated habitat patches. NMDS ordinations showed that litter-extracted arthropod compositions did not differ between the raw inoculum and rainforest habitat (Figure 4.6b, d). However, all experimental plots which received the inoculum deviated from the rainforest cluster, and the arthropod composition of all inoculated plots was similar to that within non-inoculated plots, in which few rainforest arthropods colonised (Figure 4.6a, b). ANOSIM R and P values (inoculated versus non-inoculated plots) for pitfall-trapped and litter-extracted coarse arthropods were R = -0.156, P = 0.857 and R = 0.028, P = 0.325 respectively. The same patterns were found in ordinations based on ant species composition (Figure 4.6c, d) with ANOSIM R and P values of R = 0.056, P = 0.310 and R = -0.127, P = 1.000 for pitfall traps and litter extraction respectively.

Workers of two ant species (Pheidole QM2, Ponera leae Forel) that seem to be restricted to forested areas based on existing literature (Stanisic et al. 2005) and the baseline survey (Chapter 3), were each found in a single inoculated plot. As these species were also found in the raw inoculum they may have persisted from the time the inoculum was introduced, although statistical tests were not possible. Univariate analysis showed that no variables from the inoculated plots differed significantly from those of non-inoculated plots, except

70

for the Generalised Myrmicinae, which had higher abundance in non-inoculated plots (P = 0.027).

a Coarse arthropods (Pitfall traps) b Coarse arthropods (Litter extraction)

c Ant species (Pitfall traps) d Ant species (Litter extracton)

Legend Rainforest (RF) reference Pasture (P) reference Inoculum (litter extraction only) Inoculated plot Uninoculated plot

Figure 4.6 NMDS ordination of experimental plots comparing inoculated and no-inoculation treatments at 400 m from the rainforest edge. Assemblages extracted from the raw inoculum are also shown for litter-extracted samples (b, d).

71

4.4 Discussion

There was little colonisation by ground-active rainforest-dependent taxa in any of the experimental plots beyond those closely adjacent to forest patches. This was in dramatic contrast to the predicted patterns of gradual drop off in rainforest arthropod numbers and diversity with distance. Further, this is also in contrast to the findings of other studies of recolonisation processes (Simberloff & Wilson 1969; Rey 1981; Azarbayjani et al. 1999; Sanchez & Parmenter 2002), which looked at reinvasions by aerially-dispersed fauna, following their removal from naturally occurring patches. So what prevented the invasion of my plots by rainforest fauna? I discuss five potential factors which could explain the observed results: habitat quality, plot size, habitat recognition by the fauna, colonisation speed and dispersal limitation.

Habitat quality. First, the quality of the experimental patches as habitat for rainforest-dependent arthropods may have been inadequate. The mulch lacked coarse woody debris, which provides important habitat for some forest insects (Andrew et al. 2000; Grove 2000). Further, the mulch had been sterilised with steam and then treated with herbicide, and this may have killed the arthropods’ potential food resources (e.g. prey invertebrates, bacteria, fungi). Indeed, minor differences were found in assemblage compositions between control plots (-30 m) and rainforest habitat (e.g. Figure 4.1b). Nevertheless, the control and edge plots were, in fact, colonised by taxa characteristic of rainforest (Figure 4.2c, d). This suggests that the mulch used was suitable for at least some rainforest taxa and I conclude, therefore, that habitat quality was not a major limitation to colonisation of isolated plots.

Size of the plots. The small spatial scale of the experimental design may have influenced colonisation in two ways: first, due to the scale of activity of potential colonisers and, second, due to within-plot edge effects. Some rainforest-dependent taxa (including larger-bodied, ground-active ants) may move and feed over a larger area than represented by the 9 m2 within the experimental plots. If such taxa avoid, or cannot survive in, pasture, they would not be able to persist within the isolated plots (although they could move regularly in and out of those adjacent to the forest). Even for those forest species which could potentially survive within the plots, the magnitude of the edge to area ratio may lead to intense competition with pasture species over a substantial proportion of a plot's area. In the baseline survey, many ant species were common in the matrix (pasture) 72

surrounding my experimental plots (Chapter 3) and, in this study, all plots located within pasture were colonised by ants characteristic of pasture. Empirical and experimental observations suggest that interspecific competition affects the structure of ant assemblages (Levings 1982; Morrison 2002), although this is highly variable depending on the ant assemblages studied (Gotelli & McCabe 2002; Ribas & Schoereder 2002). Within-plot edge effects may also occur as a result of changes in microclimatic conditions, such as increased mean temperatures in the plots outside the rainforest. Despite these limitations, plots adjacent to forest (0 m) were colonised by rainforest arthropods, perhaps because these plots formed ‘extensions’ of the large mass of forest habitat, reducing the effects of plot size.

Habitat recognition. Without tree cover, the experimental plots may have lacked a necessary visual, olfactory or structural stimulus to provide a proximal environmental cue for dispersing invertebrates. For ants, visual stimuli may be important factors influencing colonisation, as dispersing reproductives generally have large eyes that perceive the surrounding environment. Wilson and Hunt (1966) found that the majority of winged reproductive ants orientated their flight towards their associated habitats. Further, trees within a rainforest patch may intercept the movement of passive, wind-blown colonists, and the low height of the experimental shading structures would not have had the same effect. Grimbacher and Catterall (2007) found that canopy height was a significant predictor of the occurrence of rainforest-associated beetles in replanted rainforest plots.

Colonisation speed. The duration (nine months) of the present study may have been insufficient for assemblages of forest arthropods to arrive at, and become established, in my experimental plots. Many soil and litter arthropod taxa are wingless and, unless they are phoretic (dispersing by attaching themselves to winged organisms) or wind-dispersed (e.g. ‘ballooning’ spiderlings), they must walk to colonise new sites. A number of studies have found that wingless or short-winged arthropods colonised developing restoration areas relatively slowly compared with large-winged arthropods (Brose 2003; Moir et al. 2005). Although most ant species have winged reproductives, they may also be slow colonisers due to their eusocial life histories. Most ant colonies produce reproductives only at particular times of the year: the mated, winged queens dispersing to establish new colonies (Holldobler & Wilson 1990). Even if queens find suitable colony sites, successful establishment of a viable colony is generally rare, and may take much longer compared with recruitment rates for non-social arthropods (Holldobler & Wilson 1990). 73

Some ant species disperse by colony division which greatly reduces the risk and time taken to establish a new colony compared with a lone queen; however, dispersal capability is greatly reduced by this strategy, as all accompanying workers are wingless (Peeters & Ito 2001). Several studies have indicated that ants may be among the slowest colonisers in restored habitats in contrast to many non-social insects (Judd & Mason 1995; Williams 1997).

Dispersal limitation. The lack of rainforest arthropods in the plots at the distances of 15 m or more from the rainforest edge may indicate that dispersal by ground dwelling arthropods is either absent or limited to very short distances. Rainforest arthropods may avoid dispersing into pasture if they perceive this habitat as inhospitable. Alternatively, they may disperse randomly regardless of habitat type, but the unsuitable nature of the pasture environment prevents these rainforest arthropods reaching even the closest plots.

It is difficult to interpret which of the five factors contributed most to the low levels of colonisation observed in this study. It can be argued that poor habitat conditions (see habitat quality, size of the plots) may have had a substantial influence on the low observed colonisation rate, given that inoculation of the plots with rainforest arthropods failed to achieve significant levels of establishment. However, the failure of the inoculation was unlikely to be due to poor habitat conditions alone: other factors, such as inadequate inoculum volume and composition, and poor timing, may also explain the observed results. Consequently, the relative contributions of the various factors cannot be teased apart on the basis of my results.

The five factors broadly fall into two groups – those that are closely linked to limitations of the experimental design (habitat quality, size of the plots, habitat recognition, colonisation speed) and those that may genuinely limit arthropod colonisation in isolated habitat patches (dispersal limitation and, to a lesser extent, colonisation speed). This study used a controlled experimental design based on small-scale analogues of restoration treatments, thereby avoiding the large expense, need for large land areas, and extended time-frame of actual restoration projects. The experimental approach also circumvented the often confusing uncontrolled variability inherent in post-hoc empirical studies. In future experiments which use this approach, avoiding the effects of factors associated with the suitability of the plot condition (habitat quality, size of the plots) would require an initial demonstration of successful establishment and persistence of the target

74

arthropod taxa within the artificial habitat patches. This may potentially be achieved both by the refinement of the inoculation process as well as by improvement of the environmental conditions (e.g. addition of logs, increase in habitat size) provided by the artificial habitat patches.

While refining this approach holds some promise for future investigations, its conclusions may remain limited by the restrictions of small spatial and temporal scale. To overcome these requires controlled and replicated efforts in experimental management, over realistically large areas, involving collaboration in restoration design between researchers and practitioners, followed by longer-term monitoring of arthropod recolonisation.

75

76

5 EFFECTS OF MULCH QUALITY AND DEPTH

5.1 Introduction

With a rise in public awareness of biodiversity conservation, ecological restoration aimed at re-establishing forest ecosystems in cleared areas has become commonplace in many regions including Australia (Bennett et al. 2000; Catterall et al. 2004; Young et al. 2005). Botanical and silvicultural aspects of the development of replanted or regenerating forest vegetation have been studied extensively (e.g. Aide et al. 1996; Kooyman 1996; Ashton et al. 2001; Kanowski et al. 2003; Erskine et al. 2005). The success of ecological restoration, however, depends not only on the creation of vegetation cover, but also on the recolonization of a characteristic assemblage of fauna that is comparable with that occurring in undisturbed sites of reference forest (Majer 1989a; Scott et al. 2001; Catterall et al. 2004).

Among the large array of animals that potentially recolonise restored habitats, soil and litter arthropods are one of the most important components in terms of both abundance (biomass) and diversity, despite their inconspicuous nature (Ghilarov 1977; Andre et al. 1994). Soil and litter arthropods are also known to play a role in many ecological processes, including the decomposition of organic matter and subsequent nutrient cycling (Folgarait 1998; Hattenschwiler et al. 2005). Although the extent of arthropod diversity required to maintain such ecological processes is unknown (Bengtsson 1998; Wardle 2006), maintenance of a diverse soil and litter fauna may very well be important to the successful development of replanted vegetation in restored lands (Loreau et al. 2001).

Although the level of arthropod recolonization will ultimately be affected by many uncontrollable factors (e.g. climate and the edaphic environment, potential sources of immigrants, attributes of the surrounding matrix, natural disturbances), certain factors are, potentially, under the active control of those attempting restoration (Grove 2000; Nakamura et al. 2003; Kanowski et al. 2005; Moir et al. 2005; Redi et al. 2005; Grimbacher & Catterall 2007). For example, in the initial stages of rainforest restoration, the addition of mulch (hay, woodchips or other dead vegetation) is one such factor that may influence the survival of recolonising soil and litter arthropods. Mulching is one of the recommended management practices for replanting projects as it suppresses weed growth, retains moisture, reduces soil temperature, and may promote nutrient cycling

77

(Fimbel & Kuser 1993; Goosem & Tucker 1995; Petersen et al. 2004; Blanco-Garcia & Lindig-Cisneros 2005). Mulching may also benefit many soil and litter arthropods by providing sheltered epigeal habitats and food resources (Wardle et al. 1999a).

In rainforest restoration in tropical and subtropical Australia, a range of materials, such as hay (straw), in situ slashed weed, bagasse (sugarcane waste) bales and soaked newspaper and cardboard, have been used for mulching (Goosem & Tucker 1995; Harden et al. 2004; Big Scrub Rainforest Landcare Group 2005). The structural diversity of most of these types of mulch, however, is simpler than those of the native rainforest litter. An alternative mulch which could be used in replanting projects is forest woodchip mulch, which is sourced from managed forests (e.g. overgrown rainforest vegetation near powerlines). This type of mulch is not only similar to native litter materials found in the rainforest, but also is composed of many different plant species. Differences in both structural and nutritional quality of substrate are known to affect arthropod assemblage compositions (Hooper et al. 2000). Trait differences among various plant species should contribute to the diverse carbon resources and structural heterogeneity of litter, which may in turn allow the coexistence of a greater diversity of arthropods, than would be the case when litter is derived from one or a few plant species (Wardle 2006).

In addition to the difference in mulch quality, mulch quantity may also influence the diversity and success of soil and litter arthropod recolonisation. Increased thickness of mulch would be expected to provide better protection from desiccation and temperature fluctuations, and more food and oviposition resources, potentially facilitating colonisation by more abundant and diverse arthropods (Wardle 2006). Indeed, some authors have reported that the abundance and/or diversity of soil and litter arthropods responded positively to increasing depths of the litter layer (Kinnear 1991; Koivula et al. 1999; Chung et al. 2000; Nakamura et al. 2003; Magura et al. 2005).

Although increased depths of mulch are expected to provide hospitable habitat for a diverse array of arthropods in the context of forest restoration, little is known about the optimum amount of mulch required to facilitate recolonisation by soil and litter arthropods, especially characteristic assemblages of undisturbed reference forests. In rainforest restoration in Australia, various thickness of mulch, ranging from 2 to over 10 cm in depth, has been used or recommended to encourage the growth of replanted vegetation (Woodford 2000; Florentine & Westbrooke 2004; Big Scrub Rainforest

78

Landcare Group 2005; Cummings et al. 2005). Depths of naturally occurring litter in tropical and subtropical rainforests of Australia are reported to be similar to the lower end of mulch thickness used in actual restoration programs (mean values of rainforest litter ranges from 2.4 to 3.8 cm, King et al. 1998; Nakamura et al. 2003). Greater depths of mulch than that of naturally occurring rainforest litter may be important in providing extra protection from the severe environmental conditions that are experienced in the initial stages of rainforest restoration; however, this needs to be tested explicitly.

This chapter investigates the effect of mulch quality (hay versus woodchip mulch) and quantity (‘shallow’ versus ‘deep’) on the patterns of soil and litter arthropod colonisation in created ‘restoration’ patches. An experimental approach was adopted in which artificial plots simulating the initial stages of restored rainforest patches were created, containing varying mulch quality and quantity in areas that were previously comprised of dense pasture grasses, within a landscape which retained some remnant rainforests.

5.2 Methods

Study area The study was undertaken on the Maleny plateau, in the Sunshine Coast hinterland of eastern Australia (26° 40’- 50’ S, 152° 45’- 53’ E, elevation 350 to 530 m). Mean daily maximum and minimum temperatures in mid summer (January) are 28.9 and 18.8 ° C respectively, and 19.5 and 7.1 ° C in mid winter (July). Average annual rainfall in the region is 1973 mm, with most falling between December and April (climate data averaged over at least 50 years up to 2004 obtained from the Australian Bureau of Meteorology). It was noted that the total precipitation during the study period (1437 mm between August 2003 and April 2004) was below that of average rainfall (1633 mm).

Experimental design Five experimental sites were dispersed across a study region of around 170 km2; all within 13 km of the township of Maleny. The minimum distance between sites was 1 km; most sites were 2 to 6 km apart (Figure 2.1). Each site was comprised of cleared landscape (pasture) abutting a rainforest patch of either old regrowth (age of ca.100 years) or remnant forest which had been selectively logged until recently. None of the experimental sites shared the same rainforest remnant. Further details of study sites are provided in Section 2.1.

79

At each of the five sites, I established a series of five 3 x 3 m experimental plots, each separated by 5 m. The plots comprised replicates of two mulch types (hay or woodchip), each established at two depths: ‘shallow’ (3-5 cm) or ‘deep’ (10-15 cm). The fifth plot was a control, to which no mulch was added (see also Table 2.1, Figure 2.2). All experimental plots were situated in pasture, but were within two metres of a rainforest remnant, so that the results were not confounded by distance effects on colonisation (cf. Chapter 4).

Construction of experimental plots followed actual replanting procedures (Goosem & Tucker 1995; Big Scrub Rainforest Landcare Group 2005), based on discussions with revegetation practitioners in the study region. First, approximately 400-600 ml of broad spectrum herbicide (Roundup® Biactive™, 7.2 g/L Glyphosate) was sprayed over the entire plot area. Three weeks after the application, all visible vegetation (dead and alive) was removed by hand. Steel posts, at a height of 1.2 m, were then erected at each corner of the plot which was fenced with barbed wire to exclude stock. It should be noted that an additional test found no short- or long-term impacts of the herbicide on soil and litter arthropods inhabiting rainforest litter (Chapter 7). The plots were unshaded to simulate the initial stages of rainforest restoration. In four of the plots, sterilised hay or woodchip mulch was placed in a square area (2.5 x 2.5 m) within each quadrat. This area was bordered by wire netting (40 cm high, mesh size 1.5-2 cm) to minimise loss of mulch due to wind or disturbance by wildlife (Figure 2.3).

The main constituent of the hay was millet (Panicum spp.), one of the commonly used hay mulches in rainforest plantings in the region. Woodchip mulch was derived from lopping vegetation around powerlines in various locations within about 150 km of the study region, conducted by a local contractor of the electricity provider. Woodchip mulch comprised a mix of foliage and wood derived from rainforest and eucalypt species.

Before mulch was added to the experimental plots, it was sterilised by steam-treatment for 100 minutes. The efficacy of the treatment was tested by extracting arthropods from approximately 10 litres each of freshly steamed hay and woodchip mulch, using Tullgren funnels. The resulting extracts were checked every 4-8 hours for a total of 5 days and no live arthropods were found. Sterilised mulch was distributed to the experimental plots within seven days after steam-treatment. Before its distribution, mulch was stored beneath plastic sheets to minimise any casual arthropod colonisation.

80

The construction of the field experiment took approximately three months and was completed in August 2003. Mulch was laid down in all of the plots within one month towards the end of the construction.

Sampling methodology Arthropod sampling was carried out between 9 April and 7 May 2004 (approximately nine months after the plots were established). At each site, arthropods were collected using two methods: pitfall trapping and litter extraction. Pitfall trapping samples the relative density of active epigaeic arthropods, whereas litter extraction samples the relative abundance of sedentary arthropods that live in litter and topsoil (Majer 1997). Four 4.4 cm diameter pitfall traps (120 ml) were installed on the diagonal lines of each plot, approximately 80 cm from the centre (Figure 2.3). Vials were partly filled with 70% ethanol with a small amount of glycerol, and operated for five days. Samples from the four pitfall traps on each plot were pooled for analysis. Litter extraction was carried out by collecting 1 litre of litter and surface soil, to a depth of 1 to 2 cm, in small amounts evenly over the entire plot area (approx. 20% surface soil and 80% litter by volume). Soil and litter samples were placed in fabric bags and kept in an insulated box. Samples were placed in Tullgren funnels within 12 hours of sampling, and extracted for 4.5 days using 40 W clear light bulbs.

During sampling, care was taken to avoid cross-contamination among the experimental plots. All footwear was covered with thick polythene film, and researchers were thoroughly brush-cleaned before and after each plot was visited.

Identification of arthropods was to Order except a) Hymenoptera which were split into Formicidae and others and b) myriapods to Class. Acari, Collembola and Diptera (with the exception of litter-extracted dipterans) were not sorted due to their high abundance and ubiquitous occurrence regardless of habitat type (rainforest or pasture). Ants (Hymenoptera: Formicidae) were selected as a target group, and sorted to species. Where possible, ants were identified to described species using published taxonomic literature, otherwise they were assigned species codes. With assistance from Alan Andersen (CSIRO Sustainable Ecosystems), ant species were assigned to functional groups (viz. Climate specialists, Cryptic species, Dominant Dolichoderinae, Generalised Myrmicinae, Opportunists, Specialist predators, Subordinate Camponotini) on the basis of their

81

presumed responses to environmental stress and disturbance (see Andersen 2000a; Brown 2000).

During the arthropod sampling, soil moisture content was measured by hand-collecting approximately 50 cm3 of topsoil (up to 2-3 cm in depth) from the experimental plots of all five sites and their surrounding pasture and rainforest areas. Soil samples were kept in airtight plastic bags and the soil was weighed before and after being oven dried for 24 hours at 105 C˚. Soil surface temperature was measured by temperature loggers (HOBO® Temperature Data Logger, Onset Computer Corporation, MA) deployed at the experimental plots of one site only, and in its surrounding pasture and rainforest areas. The sensors (one per plot) were placed on the ground in plots (i.e. buried under the mulch in the experimental plots). One sensor was placed under naturally occurring litter in each of the pasture and rainforest. Temperature was recorded at 30 minute intervals from 14 to 20 April 2004 (during late summer).

Data analysis Differences between treatments were analysed in terms of three different datasets: (i) arthropods sorted to Orders/Class (referred to as ‘coarse arthropod’ hereafter), (ii) ant species and (iii) ant functional groups. Each dataset was further divided into two separate subsets comprising data sampled using pitfall traps and litter extraction. Abundances of coarse arthropods were log transformed before analysis. Abundances of ant species were scored on a seven-point scale, following Andersen et al. (2003): 1 = 1, 2 = 2-5, 3 = 6-20, 4 = 21-50, 5 = 5-100, 6 = 101-1000, 7 = >1000 individuals. Abundances within ant functional groups were expressed as the proportions of all ants recorded at each plot; proportions were arcsine transformed for analysis.

In addition to the data obtained from the present study, I also incorporated data from a preceding survey (Chapter 3), which provided baseline information on the arthropod assemblages in the pasture and rainforest habitats of the study region. That survey was carried out in the same region and in a similar season the preceding year (8 January to 6 May 2003). Pasture was sampled at sites at least 100 m away from the rainforest remnants; rainforest was sampled at sites at least 30 m from the rainforest edge except for one which was 22 m from the edge (see also Appendix 2). Arthropods were collected from three sampling points at each of the 12 sites across Maleny (N = 36), including the five sites used for the present study. Although arthropod sampling was carried out over a

82

larger area than that of the present study (each sampling point comprising a circular area of 3 m radius), sampling and sorting protocols were otherwise identical to those of the present study, so that direct comparison between the results of the baseline survey and the present study is possible.

The baseline survey was also used to identify arthropod taxa that were characteristic of either pasture or rainforest habitats (designated as ‘pasture indicators’ or ‘rainforest indicators’; see Chapter 3 for more details). Other taxa that did not appear to have any habitat preferences were classed as ‘generalists’. In the baseline survey, pasture/rainforest ‘indicators’ were split into habitat ‘specialists’ (when they were found exclusively in their preferred habitat) and ‘increasers’ (when they were found in both habitat types but occurred primarily in one); however, in this chapter they were lumped into rainforest/pasture ‘indicators’ for the sake of lucidity.

Since single taxon indicators were unreliable due to their patchy distribution, I generated additional ‘composite rainforest/pasture indices’ by summing the abundance values of all indicator taxa indicative of either rainforest (to calculate composite rainforest indices) or pasture (to calculate composite pasture indices) at each site, providing a single value which quantifies the extent to which a site is rainforest-like or pasture-like, in terms of its arthropod assemblage. Four separate composite habitat indices were obtained: for coarse arthropods and for ants, and characterising either rainforest or pasture. For the coarse arthropod dataset, abundance values of each taxon were first individually range-standardised (site-specific abundance minus minimum abundance across all sites / maximum minus minimum) to remove the effects of large differences in taxon-specific abundance, giving values between 0 and 1 for each taxon at each site. This range-standardisation procedure was not carried out for ant species, as most indicator species had a similar range in abundance score values (most ranging 0-4, with a maximum score of 5). Composite indices were calculated only if that index contained two or more of the component taxa/species, as this validates the purpose of the composite habitat indices.

Univariate analyses were carried out using two-factor crossed ANOVA with randomized complete block design to evaluate the response of variables (composite rainforest/pasture indices, total abundances, taxon richness, individual arthropod abundances, soil moisture content) to the experimental treatments: mulch quality (hay versus woodchip), mulch

83

depth (shallow versus deep) and their interaction. Individual arthropod taxa were included in the analysis only if they occurred in at least four plots from the total experimental plots used for the analyses (N = 20). Composite rainforest/pasture indices were also analysed using single-factor ANOVA with randomized complete block design across treatments involving un-mulched control, shallow and deep mulch of either woodchip or hay. This tested whether the colonisation patterns of arthropods differed in plots with either hay or woodchip mulch from those in un-mulched control plots. To permit direct comparison with pasture and rainforest reference sites, I also present arthropod data from the baseline survey obtained at the same five sites (one of the three sampling points randomly selected at a site).

Multivariate analysis was carried out to investigate responses of arthropod assemblages to the experimental treatments. I first used non-metric multi-dimensional scaling ordination (NMDS) using PRIMER v.5 software (Clarke 1993). This allowed visualisation of differences in arthropod assemblages at experimental plots in relation to pasture and rainforest reference sites. All NMDS ordinations were performed on Bray-Curtis similarity matrices, with 10 random restarts. Multivariate ANOVA (MANOVA) was carried out with PERMANOVA software (Anderson 2005) to test for responses of arthropod assemblages among the experimental treatments. This software executes MANOVA, using permutation methods, to calculate P values derived from pseudo F statistics of the distance measures (Anderson 2005). Factors tested were the same as for the ANOVA. Site variation (blocks) was accounted for in these analyses by treating ‘site’ as a covariate.

Family-wise Type I error was not controlled and the significant P threshold was set at 0.05, because the risk of Type II error was high due to the small sample size (5 replicate sites).

5.3 Results

Overall abundances A total of 7457 arthropods were sampled (5332 from pitfall traps and 2125 from litter extraction) from the field experiment. Ants were the most abundant taxon with 3437 individuals (2636 from pitfall traps and 801 from litter extraction), followed by Coleoptera with 1220 and Diplura with 554. Twenty-eight coarse arthropod taxa and 51

84

ant species were identified. Ants from litter extractions were depauperate in their abundance and species composition: 26 species were recorded and only three species were unique to this sampling method, compared with 48 species and 25 unique species from pitfall traps.

Univariate ANOVA showed that there were no effects of experimental treatments on abundance and taxon richness of pitfall-trapped coarse arthropods, whereas both abundance and taxon richness of litter-extracted coarse arthropods were significantly higher in hay than woodchip mulch (ANOVA for the effect of mulch quality: abundance, F = 6.17, P = 0.029; taxon richness, F = 11.03, P = 0.006). There were no significant effects of experimental treatments on total abundance (logged raw abundances) and species richness of either pitfall-trapped or litter-extracted ant species.

Coarse arthropods Both pitfall-trapped and litter-extracted coarse arthropods showed clear separation between pasture and rainforest reference habitats on the NMDS ordinations (Figure 5.1a, b). In pitfall-trapped coarse arthropods, data points representing the experimental plots, including un-mulched controls, were intermediate between the pasture and rainforest reference points (Figure 5.1a). MANOVA based on the same dataset showed that, although there were strong site effects, none of the experimental treatments were statistically significant (Table 5.1a). Unlike pitfall traps, the ordination based on litter-extracted coarse arthropods showed that the data points representing un-mulched controls were located within the cluster of pasture sites, and most mulched plots fell outside the clusters of both pasture and rainforest sites (Figure 5.1b). Plots with hay, especially deep hay, were closer to rainforest sites than those with woodchip mulch (Figure 5.1b). MANOVA also indicated a significant effect of mulch quality on the coarse arthropod composition (Table 5.1a).

Although both sampling methods showed a trend for higher values of composite rainforest indices in hay than in woodchip mulch (Figure 5.2a, b), this was statistically significant only for litter extraction (Table 5.2a). An additional test was carried out to determine whether the presence of mulch (either woodchip or hay) actually affected the levels of composite habitat indices compared with those found in un-mulched control plots (Table 5.3). The results showed that the levels of composite rainforest index were affected by the presence of hay; however, this was not the case for woodchip mulch: there

85

were no significant differences among plots with no mulch and shallow and deep woodchip mulch. The values of composite pasture index did not differ significantly with mulch quality or depth, and were similar in experimental and pasture reference sites (Figure 5.2c, d, Table 5.2a).

a Coarse arthropods (Pitfall traps) b Coarse arthropods (Litter extraction)

c Ant species (Pitfall traps) d Ant species (Litter extraction)

Legend

Rainforest Pasture No mulch Shallow Hay Deep Hay Shallow Woodchip Deep Woodchip

Figure 5.1 NMDS ordination of the experimental plots and rainforest and pasture site based on (a, b) coarese arthropods and (c, d) ant species. Assemblages sampled by different sampling methods (pitfall traps, litter extraction) are shown separately.

86

Table 5.1 Effects of mulch quality (hay, woodchip) and depth (‘shallow’, ‘deep’) on assemblage composition of (a) coarse arthropods, and (b) ant species. F and P values are from non-parametric MANOVA. Df for mulch quality (Quality), mulch depth (Depth), their interaction and site are 1, 1, 1, 4 respectively. Quality x Site Quality Depth Depth (covariate) F P F P F P F P a. Coarse arthropods Pitfall traps 1.23 0.288 1.80 0.101 1.35 0.247 3.56 <0.001 Litter extraction 2.99 0.009 0.57 0.775 1.37 0.226 2.09 0.002

b. Ant species Pitfall traps 1.46 0.166 0.60 0.778 1.33 0.251 3.63 <0.001 Litter extraction 0.62 0.768 1.20 0.055 1.34 0.232 1.44 0.072 Significant P values (< 0.05) are shown in bold.

a Composite rainforest index (Pitfall traps) b Composite rainforest index (Litter extraction) 5 6

4 5

4 3 3 2 2 Index level level Index Index level level Index 1 1

0 0 ShallowS HM DDeep HM Shallow S WM Deep D WM None No Pasture P Rain- RF ShallowS HM DDeep HM Shallow S WM Deep D WM None No Pasture P Rain- RF Hay Hay Wood Wood Mulch forest Hay Hay Wood Wood Mulch forest

c Composite pasture index (Pitfall traps) d Composite pasture index (Litter extraction) 2 1

1.5

1 0.5 Index level level Index Index level level Index 0.5

0 0 ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF Hay Hay Wood Wood Mulch forest Hay Hay Wood Wood Mulch forest

Figure 5.2 Effects of mulch quality and depth on composite rainforest and pasture indices of coarse arthropods (mean index level, SE). Values for pasture and rainforest reference habitats (sampled the previous year) are also shown. The plots included in the analysis of two-factor crossed ANOVA (Table 5.2) were represented by closed bars. Wood = woodchip mulch.

87

88

Table 5.2 Effects of mulch quality (hay, woodchip) and depth (‘shallow’, ‘deep’) on levels of composite rainforest and pasture indices for (a) coarse arthropods, and (b) ant species. F and P values are from two-factor crossed ANOVA with randomized complete block design. Df for mulch quality (Quality), mulch depth (Depth), their interaction and site are 1, 1, 1, 4 respectively. Some composite indices were not calculated, as less than 2 component species (T < 2) were found. Quality x Quality Depth Depth Site (block) F P F P F P F P Higher in a. Coarse arthropods Composite rainforest index (Pitfall traps) 3.34 0.093 1.17 0.301 0.89 0.364 3.57 0.039 - Composite rainforest index (Litter extraction) 9.10 0.011 1.14 0.306 3.14 0.102 3.00 0.062 Hay

Composite pasture index (Pitfall traps) 0.04 0.841 0.02 0.885 1.31 0.275 4.27 0.022 - Composite pasture index (Litter extraction) ------

b. Ant species Composite rainforest index (Pitfall traps) 8.70 0.012 15.06 0.002 8.70 0.012 2.83 0.073 Shallow hay Composite rainforest index (Litter extraction) ------

Composite pasture index (Pitfall traps) 2.01 0.181 0.06 0.806 0.17 0.684 3.37 0.046 - Composite pasture index (Litter extraction) 2.72 0.125 0.02 0.883 0.20 0.661 1.72 0.210 - Significant P values (< 0.05) are shown in bold.

Table 5.3 Effects of mulch depth involving the un-mulched control, ‘shallow’ mulch and ‘deep’ mulch treatments, on levels of composite rainforest and pasture indices for (a) coarse arthropods, and (b) ant species. Analysis was carried out separately for woodchip and hay mulch. F and P values are from single-factor ANOVA with randomized complete block design. Df for mulch depth (Depth) and site are 2 and 4 respectively. Some composite indices were not calculated, as less than 2 component species (T < 2) were found. Woodchip Hay Depth Site (block) Depth Site (block) F P F P F P F P a. Coarse arthropods Composite rainforest index (Pitfall traps) 0.15 0.861 2.43 0.133 4.10 0.060 3.88 0.049 Composite rainforest index (Litter extraction) 1.11 0.377 4.02 0.045 6.09 0.025 1.88 0.208

Composite pasture index (Pitfall traps) 1.36 0.311 7.27 0.009 2.19 0.175 4.45 0.035 Composite pasture index (Litter extraction) ------

b. Ant species Composite rainforest index (Pitfall traps) 0.38 0.699 2.75 0.104 25.12 <0.001 6.03 0.015 Composite rainforest index (Litter extraction) ------

Composite pasture index (Pitfall traps) 0.56 0.594 1.19 0.387 0.03 0.972 4.04 0.044 Composite pasture index (Litter extraction) 1.41 0.300 1.24 0.366 0.18 0.841 1.99 0.189 Significant P values (< 0.05) are shown in bold.

89

Among the pitfall-trapped coarse arthropods, 11 taxa were rainforest indicators (Archaeognatha, Blattodea, Coleoptera, Dermaptera, Diplopoda, Diplura, Heteroptera, Isopoda, Opilionida, Pseudoscorpionida, Psocoptera) and three taxa were pasture indicators (Araneae, Homoptera, Orthoptera). Among litter-extracted arthropods, there were 14 rainforest indicators (Amphipoda, Blattodea, Chilopoda, Coleoptera, Dermaptera, Diplopoda, Diplura, Formicidae, Heteroptera, Isopoda, ‘other Hymenoptera’, Pauropoda, Pseudoscorpionida, Symphyla) and one pasture indicator (Orthoptera). The remaining taxa were classed as ‘generalists’. Statistical tests were possible on eight pitfall-trapped rainforest indicators, eight litter-extracted rainforest indicators, and all pitfall-trapped and litter-extracted pasture indicators. The experimental treatments were statistically significant for a number of taxa (Table 5.4). In pitfall-trapped coarse arthropods, two pasture indicators (Figure 5.3a, b) and three ‘generalists’ (Figure 5.3c-e) responded to the experimental treatments, but their responses were inconsistent within their habitat indicator groups (Table 5.4). Litter-extracted coarse arthropods had three rainforest indicators (Figure 5.4a-c) and one ‘generalist’ (Figure 5.4d) that responded positively to either hay or deep hay (Table 5.4). These results were consistent with those obtained from higher-order analysis.

90

Table 5.4 Effects of mulch quality (hay, woodchip) and depth (‘shallow’, ‘deep’) on abundances of pitfall-trapped and litter-extracted coarse arthropods. Only taxa that had significant response to the treatments are shown. Their habitat indicator category (rainforest indicator, pasture indicator, ‘generalist’) is shown in parenthesis. F and P values are from two-factor crossed ANOVA with randomized complete block design. Df for mulch quality (Quality), mulch depth (Depth), their interaction and site are 1, 1, 1, 4 respectively. Quality x Quality Depth Depth Site (block) F P F P F P F P Higher in Pitfall traps Araneae (Pasture indicator) 1.62 0.227 0.89 0.364 5.53 0.037 3.82 0.032 Deep woodchip Homoptera (Pasture indicator) 5.79 0.033 0.01 0.906 0.33 0.576 3.78 0.032 Hay Amphipoda (‘Generalist’) 3.15 0.101 6.44 0.026 1.32 0.272 6.95 0.004 Deep Chilopoda (‘Generalist’) 0.04 0.854 6.75 0.023 0.54 0.477 0.23 0.916 Deep Other Hymenoptera (‘Generalist’) 6.90 0.022 0.90 0.361 1.21 0.293 8.97 0.001 Hay

Litter extraction Coleoptera (Rainforest indicator) 9.84 0.009 0.63 0.443 5.07 0.044 0.68 0.619 Deep hay Diplura (Rainforest indicator) 8.04 0.015 3.69 0.079 1.62 0.227 8.34 0.002 Hay Isopoda (Rainforest indicator) 9.35 0.010 1.90 0.193 6.90 0.022 0.95 0.470 Deep hay Psocoptera (‘Generalist’) 8.44 0.013 1.22 0.291 1.22 0.291 0.61 0.663 Hay Significant P values (< 0.05) are shown in bold.

91

a Araneae (Pasture indicator) b Homoptera (Pasture indicator) 25 60

20

40 15

10 Abundance 20 Abundance 5

0 0 Shallow Deep Shallow Deep No Pasture Rain- S HM D HM S WM D WM None P RF ShallowS HM Deep D HM Shallow S WM DDeep WM NoneNo Pasture P Rain- RF Hay Hay Wood Wood Mulch forest Hay Hay Wood Wood Mulch forest c Amphipoda (‘Generalist’) d Chilopoda (‘Generalist’) 6 10

5 8

4 6 3 Abundance Abundance Abundance 4 2

1 2

0 0 ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF Hay Hay Wood Wood Mulch forest Hay Hay Wood Wood Mulch forest

e Other Hymenoptera (‘Generalist’)

12

8 Abundance Abundance 4

0 ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF Hay Hay Wood Wood Mulch forest

Figure 5.3 Effects of mulch quality and depth on abundances (mean, SE) of pitfall-trapped coarse arthropod taxa that showed statistically significant differences. Values for pasture and rainforest reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars. Wood = woodchip mulch.

92

a Coleoptera (Rainforest indicator) b Diplura (Rainforest indicator)

50 80

40 60

30 40 20 Abundance Abundance Abundance 20 10

0 0 ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF ShallowS HM Deep D HM Shallow S WM Deep D WM No None Pasture P Rain- RF Hay Hay Wood Wood Mulch forest Hay Hay Wood Wood Mulch forest

c Isopoda (Rainforest indicator) d Psocoptera (‘Generalist’) 80 4

60 3

40 2

20 1 Abundance Abundance

0 0 ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF Hay Hay Wood Wood Mulch forest Hay Hay Wood Wood Mulch forest

Figure 5.4 Effects of mulch quality and depth on abundances (mean, SE) of litter-extracted coarse arthropod taxa that showed statistically significant differences. Values for pasture and rainforest reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars. Wood = woodchip mulch.

Ants Multivariate analyses of ant species assemblages from both pitfall traps and litter extraction showed no obvious patterns among the experimental plots. On the ordinations, all the experimental plots clustered between the pasture and rainforest reference habitats (Figure 5.1c, d). MANOVA showed that there were no significant treatment effects (except site effects) on ant species assemblages (Table 5.1b).

Levels of composite rainforest indices of pitfall-trapped ant species were considerably and significantly greater in plots with shallow hay, almost reaching similar levels to those found in the rainforest reference habitat (Figure 5.5a, Table 5.2b). For composite pasture indices, all experimental plots had similar index levels to those found in un-mulched control and pasture (Figure 5.5c). Composite indices from litter-extracted ants did not

93

show any clear patterns as few rainforest or pasture habitat indicators were sampled (Figure 5.5b, d, Table 5.2b).

a Composite rainforest index (Pitfall traps) b Composite rainforest index (Litter extraction) 8 8

6 6

4 4 Index level level Index Index level level Index 2 2

0 0 ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF Hay Hay Wood Wood Mulch forest Hay Hay Wood Wood Mulch forest

c Composite pasture index (Pitfall traps) d Composite pasture index (Litter extraction) 10 3

8

2 6

Index level level Index 4 Index level level Index 1 2

0 0 ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF Hay Hay Wood Wood Mulch forest Hay Hay Wood Wood Mulch forest

Figure 5.5 Effects of mulch quality and depth on composite rainforest and pasture indices of ant species (mean index level, SE). Values for pasture and rainforest reference habitats (sampled the previous year) are also shown. The plots included in the analysis of two-factor crossed ANOVA (Table 5.2) were represented by closed bars. Wood = woodchip mulch.

Among the pitfall-trapped ant species, seven taxa were rainforest indicators (Anonychomyrma QM3, Leptomyrmex erythrocephalus rufithorax, Notoncus capitatus, Pachycondyla QM2, Pheidole QM1, Pheidole QM2, Pheidole sp.2) and five taxa were pasture indicators (Cardiocondyla nuda, Carebara QM1, Paratrechina QM2, Pheidole QM3, Rhytidoponera metallica). Among litter-extracted ant species, there were only one rainforest indicator (Hypoponera sp.1) and two pasture indicators (Pheidole QM3, Carebara QM1). The remaining species were classed as ‘generalists’. None of the rainforest or pasture indicators showed significant responses to the experimental treatments, although two pitfall-trapped rainforest indicators showed trends for increased abundance in shallow hay (Figure 5.6a, b). One pitfall-trapped pasture indicator, Rhytidoponera metallica, was abundant in most experimental plots as well as pasture, but was absent from rainforest sites (Figure 5.6c). Occurrences of other indicators were too

94

few to interpret the patterns. Effects of the experimental treatments were statistically significant for two ‘generalist’ ant species from pitfall traps (Figure 5.6d, e), which responded positively to woodchip mulch (ANOVA for the effect of mulch quality: F = 8.58, P = 0.013; F = 6.40, P = 0.026 for Rhytidoponera victoriae and Solenopsis QM1 respectively).

a Anonychomyrma QM3 (Rainforest indicator) b Pheidole QM1 (Rainforest indicator) 2.5 2.5

2 2

1.5 1.5

1 1 Abundance scale 0.5 scale Abundance 0.5

0 0 Shallow Deep Shallow Deep No Pasture Rain- S HM D HM S WM D WM None P RF ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF Hay Hay Wood Wood Mulch forest Hay Hay Wood Wood Mulch forest

c Rhytidoponera metallica (Pasture indicator) d Rhytidoponera victoriae (‘Generalist’) 4 3

3

2 2

1

Abundance scale 1 Abundance scale

0 0 ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF ShallowS HM Deep D HM Shallow S WM Deep D WM No None Pasture P Rain- RF Hay Hay Wood Wood Mulch forest Hay Hay Wood Wood Mulch forest

e Solenopsis QM1 (‘Generalist’) 4

3

2

1 Abundance scale

0 ShallowS HM Deep D HM Shallow S WM Deep D WM NoneNo Pasture P Rain- RF Hay Hay Wood Wood Mulch forest

Figure 5.6 Effects of mulch quality and depth on abundance scales (mean, SE) of selected ant species from pitfall traps. Values for pasture and rainforest reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars. Wood = woodchip mulch.

95

Over 85% of the total ants belonged to Opportunists (1375 individuals), Generalised Myrmicinae (999) or Cryptic species (651). Dominant Dolichoderinae was much less abundant with 188 individuals, and Subordinate Camponotini was absent. Amongst ant functional groups, only Opportunists from pitfall traps showed significant responses to experimental treatments: their relative abundance was higher in deep hay (ANOVA for the effect of interaction: F = 7.46, P = 0.018). Other functional groups showed no notable responses to mulch quality and quantity.

Soil moisture content and temperature The mean soil moisture content was higher in pasture than in rainforest (Figure 5.7). Although the mulched plots retained more moisture compared with the un-mulched controls, the soil moisture contents were considerably lower than those found in either rainforest or pasture (Figure 5.7). Among mulched plots, soil moisture was higher in those with woodchip mulch (ANOVA for the effect of mulch quality: F = 21.89, P = 0.001), but was not significantly influenced by mulch depth (F = 0.04, P = 0.837).

60

50

40

30

20

10 Soil moisture content (%) (%) content moisture Soil 0 ShallowSH Deep DH Shallow SW Deep DW NoneNo Pasture Pasture Rain- RF Hay Hay Wood Wood Mulch forest

Figure 5.7 Effects of mulch quality and depth on soil moisture contents (mean, SE) across the experimental plots. Values for pasture and rainforest reference habitats are also shown. All samples were taken in 2004. The plots included in the statistical analysis were represented by closed bars. Wood = woodchip mulch.

96

Compared with rainforest, pasture had higher average temperature, and the coefficient of variation was almost twice as high (Figure 5.8a). Hay plots showed lower average temperatures than woodchip mulch, although average temperatures of all the mulched plots were higher than pasture (Figure 5.8b). Among the mulched plots, deep hay had the smallest temperature fluctuations.

a Mean CV 35 No mulch 20.84 23.92 Pasture 20.02 6.23 18.28 3.50 ) Rainforest ˚ 30

25

Temperature (C Temperature 20

15 0 20 40 60 80 100 120 140

b Mean CV 35 Shallow hay 20.68 8.57 Deep hay 21.62 5.38 )

˚ Shallow woodchip 23.21 7.66 30 Deep woodchip 22.15 8.43

25

Temperature (C Temperature 20

15 0 20 40 60 80 100 120 140 Time lapsed (hour)

Figure 5.8 Temperature fluctuations over five days in late summer (April) recorded at: (a) un-mulched control plots, pasture and rainforest reference habitats, and (b) plots with mulching treatments, with mean values and coefficient of variations (CV). Data from one site only (E3, D3).

97

5.4 Discussion

Experimental approach in restoration studies Despite an increasing number of restoration studies investigating recolonisation by various faunal groups, the majority have been based on empirical observation of actual restoration sites (see also Section 1.3.3). Their findings, therefore, are often confounded by other factors (Michener 1997; Catterall et al. 2004). For example, time since restoration commenced and plant species composition and biomass are likely to vary together, confounding the effects of litter quality and quantity in the field. The experimental approach used in this study removes such potentially confounding factors. For soil and litter arthropods targeted in the present study, a small-scale field experiment was expected to be adequate. This is because many of these animals are small and sedentary (van Straalen 1997, 1998), requiring relatively small area for their survival and reproduction. Further, they exhibit rapid population growth and short generation turnover, thereby responding to the changes in micro-habitat conditions more sensitively and quickly than other large-bodied organisms, such as vertebrates (Kremen et al. 1993). Indeed, despite the small patch size and short duration of the field experiment, arthropods colonised the experimental plots and exhibited various responses to the mulch treatments.

Effect of mulch on soil and litter arthropods While there is broad consensus regarding the positive relationship between diversity of above-ground arthropods and plants (Hunter & Price 1992; Lewinsohn et al. 2005b), it is not clear whether the same relationship generally applies to the diversity of soil and litter arthropods and plants that supply litter because the results of empirical studies are often contradictory (e.g. Crisp et al. 1998; Ferrier et al. 1999; Wardle et al. 1999b; Proctor et al. 2003). In all but one of the four manipulative field experiments conducted to date, soil and litter arthropods consistently showed a trend for increased species richness and abundances with increased species composition and structural complexity of litter (Kaneko & Salamanca 1999; Hansen 2000; Armbrecht et al. 2004; Wardle et al. 2006). My study, however, did not find such a relationship: hay mulch, despite its simple composition, performed better in facilitating colonisation by arthropod assemblages characteristic of rainforest. Surprisingly, levels of colonisation by rainforest arthropods in

98

plots with woodchip mulch were no greater than un-mulched control plots, regardless of its depth (Table 5.3).

While previous field experiments were conducted in forest environments, my study simulated newly restored sites that received little protection from their surrounding vegetation; thereby colonising arthropods are exposed to severe microclimatic conditions unless other means of protection (such as mulch) are provided. Many forest-associated arthropods were associated with cool and moist environments (Levings & Windsor 1984; Dindal 1990; King et al. 1998), suggesting that differences in degree of insulation and moisture retention provided by different types of mulch may well be important. My temperature measurements showed lower average temperatures in hay than those of woodchip mulch, due probably to its better insulation capacity and/or lighter colour. Soil moisture content did not explain the patterns of arthropod assemblages because the values of moisture content was higher under woodchip mulch than that of hay, and all of the experimental plots had lower values than in rainforest and pasture reference sites. Uetz (1979) found that contribution of different microhabitat conditions, such as temperature and moisture contents, varies significantly depending on seasonal variation of climatic conditions. As I collected arthropods in the hot-wet season, mitigation of extreme temperature fluctuations was more important, favouring hay as suitable habitat for rainforest arthropods. Soil moisture would have been more important had I collected arthropods in the cool-dry season, which may have generated different patterns in arthropod colonisation. However, arthropod activity and abundance may also be low at all sites in winter.

One may argue that the differences in temperature do not explain colonisation patterns of rainforest arthropods because average temperatures of experimental plots were all greater than that of pasture. It should be noted, however, that pasture sites were subject to livestock grazing and were located at least 100 m away from the rainforest edge, whereas all experimental plots eliminated livestock grazing and were located adjacent to the rainforest edge. These differences were reflected in control plots whose levels of composite rainforest index were apparently intermediate between those of pasture sites and mulched experimental plots, despite substantial temperature fluctuations that the control plots received (see Figures 5.2a, b, 5.5a). Therefore, alleviated temperature regime may facilitate colonization by rainforest arthropods, when other overriding factors (i.e. livestock grazing and distance effect) were controlled. 99

An alternative (but not mutually exclusive) to the above explanation is that hay may have provided more food resources than that from woodchip mulch. Compared with hay, woodchip mulch which composed of less foliar and more woody materials generally has higher lignin content (Tian et al. 1993) – one of the major plant tissue that is recalcitrant to decomposition (Berg & McClaugherty 2003). It is thereby suggested that relatively small amount of foliage materials in woodchip mulch may have been quickly depleted, leaving little food resources for arthropods that depend on them. Although decomposing woody materials are known to provide important food sources for many forest dwelling arthropods (e.g. Grove 2000), only a small amount of the woody component of this mulch may have decomposed within the study duration. Although I did not measure the decomposition rate of the mulch used in the present study, quickly decomposing hay may have provided more food resources to decomposer arthropods (Mando et al. 1999) that were abundant in rainforest habitats (Chapter 3). Indeed, some decomposer arthropods (detritivores and fungivores, see litter-extracted Diplura, Isopoda, Psocoptera in Figure 5.4) showed increased abundances in the hay treatment.

An increase in these decomposer arthropods as well as some herbivores (e.g. Homoptera), however, did not appear to be related to the abundances of predatory arthropods that feed on them. Different responses of the predatory taxa (viz. Araneae, Chilopoda, other Hymenoptera, see Figure 5.3) to the experimental treatments indicate that neither habitat complexity nor their potential prey abundances (or factors that increased/decreased prey abundances) can be used to generalise habitat preferences of predatory arthropods. Using a meta-analytical approach, Lengellotto and Denno (2004) concluded that abundances of predatory arthropods responded positively to increased habitat complexity. However, nearly half of the studies that were used in their analyses investigated one particular group of predators (Araneae), and much less attention was paid to other groups (e.g. predatory Coleoptera, Formicidae). A more extensive suite of predatory taxa is needed to reconcile the generality of this relationship.

Responses of rainforest-associated ant species to the mulch treatments were less predictable than the patterns found for coarse arthropods. Disproportional increase of the composite rainforest index in shallow hay may be attributable to the bias in ant ‘trappability’ (Melbourne 1999), as only pitfall-trapped ants responded to the experimental treatments. Melbourne (1999) demonstrated that many ant species caught by pitfall trapping responded to different complexities of the surrounding ground cover, 100

and their abundances and occurrences were often reduced in areas with ample litter. It is, however, not clear whether trappability alone explains the result of the present study, as only certain species of rainforest-associated ants responded positively to shallow mulch. Positive responses of rainforest ants in shallow hay may be explained by the lowest average temperature experienced in this treatment. Further, similar depth of mulch (ca. 3 cm) to that of naturally occurring rainforest litter (King et al. 1998; Nakamura et al. 2003) may have been perceived by rainforest ants as a suitable foraging habitat.

Unlike pitfall-trapped ants, rainforest ants from litter extraction did not show any notable patterns and the levels of the composite rainforest index were very low across the experimental treatments. As pitfall traps generally collect actively moving epigaeic ants, increased levels of composite rainforest index by pitfall trapping alone may suggest that the captured rainforest ants only used shallow hay during foraging movements, when the conditions were more suitable during certain times of the day. It was noted that the temperatures in shallow hay were considerably lower during the night. Certain rainforest species of actively moving ants may therefore forage in the plots only at night, and may retreat into rainforest when temperatures rise during the day. Un-mulched control plots also experienced relatively low temperatures during the night; however, they may have been avoided by rainforest ants if the absence of mulch was perceived as unsuitable for foraging.

Litter extraction often collects many unique species of cryptic ants, which forage and nest in topsoil and litter of forest floor (Andersen 1995a; Bestelmeyer et al. 2000). While the baseline survey identified many cryptic species of ‘rainforest indicators’ (e.g. Discothyrea, Hypoponera, Strumigenys spp., Chapter 3), few of them were collected in the experimental plots. Although the absence of these species may simply indicate lack of sampling intensity, it is perhaps more likely to indicate that the addition of mulch alone did not create suitable habitats for these species. Supporting this, a study carried out in rainforest restoration of former mining sites in Brazil (Majer 1996) suggested that composition of litter-extracted ant species may be more sensitive than that of other epigaeic ants (i.e. pitfall-trapped ants) in responding to habitat conditions of restored lands.

Although the results supported the use of hay over woodchip mulch as habitat for rainforest arthropods, neither hay nor woodchip impeded colonisation by arthropods

101

characteristic of pasture. The provision of mulch, irrespective of mulch quality and quantity, appeared to be unimportant for colonisation by pasture-associated arthropods at least within these experimental plots. If one aim of restoration is to exclude recolonisation by arthropods characteristic of the habitat matrix, other factors associated with the development of restored rainforest may be important. Another study of the same field experiment (Chapter 6) found that pasture-associated arthropods (including ant species) responded negatively to increased levels of canopy cover, suggesting that the establishment of canopy cover may be essential in impeding colonisation by pasture-associated soil and litter arthropods.

Management implications and future directions In rainforest restoration of the study region, hay has been one of the most commonly used mulches because of its cost effectiveness and ease of handling (Big Scrub Rainforest Landcare Group 2005). Within the initial stages of rainforest restoration (up to nine months), hay appeared to perform better than woodchip mulch in facilitating colonisation by rainforest-associated soil and litter arthropods. The most suitable depth of mulch to encourage rainforest arthropod colonisation was, however, not clear as two levels of taxonomic resolution (i.e. coarse arthropods and ant species) responded differently to mulch depth.

This study investigated initial stages of rainforest restoration that spanned up to nine months during a summer period. Development of the soil and litter arthropod assemblages after this period of time, however, was not followed. A study by Tian et al. (1993) showed that temperatures and soil moisture content of different types of mulch changed over time under field conditions in the humid tropics, affecting compositions of assemblages of soil and litter arthropods. Further studies are needed to investigate how long hay may support rainforest-associated arthropods, and whether restored sites subsequently require additional inputs of mulch until litter is supplied naturally from developing vegetation. It is also important to investigate whether woodchip mulch accommodates rainforest arthropods as decomposition of woody materials progresses.

102

6 EFFECTS OF SHADING AND MULCH DEPTH

6.1 Introduction Despite increasing concerns from both the scientific community and the general public (e.g. Sala et al. 2000), the rate of deforestation between 2000 and 2005 remained alarmingly high in many of the world’s tropical regions (FAO 2006). Australian tropical and subtropical rainforests represent one of the few cases in which deforestation has all but ceased, and increasing effort is being invested in restoring rainforest on cleared landscapes, often with the aims of recovering biota characteristic of pre-disturbed habitats (Erskine 2002; Tucker et al. 2004; Catterall & Harrison 2006).

In order to develop more effective restoration practices, scientific and practical knowledge must be gathered through studying the recolonisation of restored habitat by biota (Bradshaw 1987; Brown & Lugo 1994). Along with the studies on development of replanted vegetation, it is equally important to study development of faunal assemblages potentially colonising restored habitats. Studies of faunal colonisation of restoration sites, however, predominantly come from former minesites within sclerophyllous or temperate forests (Majer 1989a; Underwood & Fisher 2006, see also Table 1.2). In contrast, our understanding of faunal recolonisation of restored rainforest on former pasture is still in its infancy (Catterall et al. 2007).

The last decade has seen a rapidly increasing number of studies of the restoration of tropical and subtropical rainforests, investigating recolonisation by various faunal groups including small mammals (Tucker 2001), birds (Passell 2000; Catterall et al. 2004; Jansen 2005; Kanowski et al. 2005), reptiles (Kanowski et al. 2006) and arthropods (Majer 1992, 1996; Jansen 1997; Nakamura et al. 2003; Proctor et al. 2003; Catterall et al. 2004; Kanowski et al. 2005; Grimbacher & Catterall 2007; Grimbacher et al. 2007). Together with spatio-temporal aspects of rainforest restoration (e.g. isolation, age of restoration), much attention has been paid to biological (e.g. plant species richness, interaction with other existing or recolonising biota) as well as physical (e.g. plant structural complexity, canopy cover, litter depth, woody debris) factors that may potentially influence the recolonisation patterns of rainforest fauna.

Among these factors investigated, canopy closure is considered by a number of restoration studies as a ‘keystone structure’ (sensu Tews et al. 2004) in facilitating the

103

development of fauna, including soil and litter arthropods, especially during early stages of rainforest restoration (Jansen 1997; Nakamura et al. 2003; Kanowski et al. 2006; Catterall et al. 2007; Grimbacher et al. 2007). A closed canopy provides a shaded forest floor, which is in turn associated with increased moisture content and reduced temperature fluctuations in the soil and litter microhabitat (Neumann 1973). These factors are important determinants of the regulating diversity and abundance of soil and litter arthropod assemblages (Remmert 1981; Parmenter et al. 1989; Chikoski et al. 2006; Entling et al. 2007).

The level of shading achieved during rainforest restoration therefore has important implications for the development of rainforest-like arthropod assemblages, as different reforestation techniques often achieve different degrees of canopy cover. A number of authors have advocated the use of timber plantations to catalyse rainforest reforestation because plantations can also yield economic returns (Lugo 1997; Erskine et al. 2005). However, the density of planted trees in plantations is generally sparse (400-1000 stems/ha) compared with ecological restoration where a diverse array of rainforest plants are planted at densities of several thousand stems/ha (Catterall & Harrison 2006). Kanowski et al. (2003) reported that, in tropical and subtropical regions, timber plantations achieved much lower levels of canopy cover (25-60%) than ecological restoration plantings (75-80%) and intact rainforest (93-95%), at least during the earlier stages of reforestation (5-22 years since establishment).

If increased insolation due to an open canopy is found to inhibit colonisation and establishment of rainforest fauna in plantations, it may be desirable to manipulate their structure in other ways to promote their habitat value for fauna. In particular, increasing the amount of litter may help offset the more open canopy of plantations, by providing better insulation, and more habitat and food resources for recolonising soil and litter arthropods (Majer et al. 1984; Greenslade & Majer 1993; Koivula et al. 1999; Nakamura et al. 2003). This may be achieved through the addition of a layer of organic mulch (such as woodchip or hay) during the early stages of restoration. However, whether this strategy would be effective is currently unknown, as the effects of shading and litter depth on colonisation patterns of soil and litter arthropods have not been systematically studied in restored rainforests.

104

In this chapter, I investigated the effects of shading and mulch depth on the development of soil and litter arthropods in experimentally-created patches of potential habitat, within a rainforest landscape now dominated by pasture. The experimental approach enabled me to systematically test the focal factors without interactions with extraneous factors, such as litter composition, habitat area and proximity to the nearest rainforests, which are inherently variable in studies of actual restoration programs. I test the hypothesis that the abundance and diversity of rainforest-associated arthropods in restored patches will increase with increased shading and that an opposite pattern will occur for pasture-associated arthropods. Further, I test whether increased mulch depth will compensate for the effects of reduced shading, thereby increasing the colonisation of open-canopied sites by rainforest-associated arthropods.

6.2 Methods

Study area The study was undertaken on the Maleny plateau, in the Sunshine Coast hinterland of eastern Australia (26° 40’- 50’ S, 152° 45’- 53’ E, elevation 350 to 530 m). Mean daily maximum and minimum temperatures in mid summer (January) are 28.9 and 18.8 ° C respectively, and 19.5 and 7.1 ° C in mid winter (July). Average annual rainfall in the region is 1851 mm, with most falling between December and April (climate data averaged over at least 50 years up to 2004, obtained from the Australian Bureau of Meteorology). It was noted that the total precipitation during the study period (1437 mm between August 2003 and April 2004) was below the average rainfall for those months (1633 mm).

Five experimental sites were dispersed across a study region of approximately 170 km2; all within 13 km of the township of Maleny. The minimum distance between sites was 1 km; most sites were > 2 km apart (Figure 2.1). Each site comprised an area of pasture abutting a rainforest remnant. The pasture at four of the sites was heavily grazed by cattle and/or horses, and lightly grazed at the other (see also Appendix 2). Adjacent rainforest remnants were either old regrowth (age of ca.100 years) or remnants that had been selectively logged until recently. Edges of the rainforest remnants were fenced to exclude livestock. None of the sites shared the same remnant rainforest. Further details of study sites are provided in Section 2.1.

105

Experimental design At each site, I established a series of experimental plots (3 x 3 m quadrats) that simulated conditions experienced by soil and litter arthropods within areas of rainforest restoration. All experimental plots were situated in pasture, but were within two metres of a rainforest remnant, so that the results were not confounded by distance effects on colonisation (see Chapter 4). Plots were at least 5 m apart.

Construction of experimental plots simulated actual replanting procedures (Kanowski et al. 2003; Big Scrub Rainforest Landcare Group 2005; Catterall et al. 2007), based on discussions with revegetation practitioners in the study region. Experimental plots were first sprayed with broad spectrum herbicide (Roundup® Biactive™, 7.2 g/L Glyphosate). Three weeks after the application, all visible vegetation (dead and alive) was removed by hand. A companion study found no short- or long-term impacts of the herbicide on soil and litter arthropods inhabiting rainforest litter (Chapter 7). Steel posts, at a height of 1.2 m, were then erected at each corner of the plot which was fenced with barbed wire to exclude stock (Figure 2.3).

To simulate shading effects produced by a developing tree canopy, the plots were either unshaded or covered with Sarlon® shadecloth rated at either 50% or 90% protection from insolation. Shadecloth was placed over the top of the quadrat and 20-40 cm down each side. The shadecloth was cut in 15-20 places with a slit length of 10-15 cm to permit sunflecks and throughfall of rain.

To provide a potential habitat for soil and litter arthropods, a mulch of woodchip and leaf material was placed at two different depths (‘shallow’, 3-5 cm; ‘deep’, 10-15 cm) in a square quadrat (2.5 x 2.5 m) established inside each plot. The area was bordered by wire netting (40 cm high, hexagonal mesh size of 1.5 cm [maximum height] x 2 cm [maximum width]) to minimise loss of mulch due to wind or disturbance by wildlife. The mulch material was derived from lopping vegetation around powerlines in various locations within about 150 km of the study region, conducted by a local contractor of the electricity provider. Woodchip mulch comprised a mix of foliage and wood, derived from rainforest and eucalypt species. Before mulch was added to the experimental plots, it was sterilised by steam-treatment for 100 minutes. The efficacy of the treatment was tested by extracting the arthropods from approximately 10 litres of freshly-steamed woodchip mulch, using Tullgren funnels (see below). The resulting extracts were checked every 4-8

106

hours for a total of 5 days and no live arthropods were found. Sterilised mulch was distributed to the experimental plots within seven days after steam-treatment. Before its distribution, mulch was stored beneath plastic sheets to minimise any casual arthropod colonisation.

A total of seven plots were constructed at each of the five replicated sites. Six of these plots were experimental treatments, with three levels of shading (0%, 50%, 90%) and two levels of mulch depth (‘shallow’, ‘deep’), and the seventh was a control plot which received herbicide treatment and vegetation removal but no shadecloth or mulch (see also Table 2.1, Figure 2.2). Construction of the field experiment took place between May and August, 2003.

Sampling methodology Sampling was carried out between 9 April and 7 May 2004 (approximately nine months after the plots were established). At each site, the relative density of arthropod taxa was assessed using two methods: pitfall trapping and litter extraction. Four pitfall traps were installed on the diagonal lines of each plot, approximately 80 cm from the centre (Figure 2.3). Each trap was a 120 ml plastic vial (44 mm in diameter), buried in the ground with the lip flush with the surface. Vials were then filled with 70 to 80 ml of 70% ethanol with a small amount of glycerol. Pitfall traps were operated for five days. Prior to data analyses, samples from the four pitfall traps were pooled. Litter extraction was carried out by collecting 1 litre of litter and surface soil (to a depth of 1 to 2 cm) in small amounts evenly over the entire plot area. The same proportion of surface soil relative to litter was collected (approx. 20% surface soil and 80% litter by volume). Samples were placed in Tullgren funnels within 12 hours of sampling, and extracted for 4.5 days using 40 W clear light bulbs.

During sampling, care was taken to avoid cross-contamination among the experimental plots. All footwear was covered with thick polythene film and researchers were thoroughly brush-cleaned before and after each plot was visited.

All insects and arachnids were sorted to Order except a) Hymenoptera which were split into Formicidae and others; and b) myriapods, which were sorted to Class. Acari, Collembola and Diptera (with the exception of litter-extracted dipterans) were not sorted due to their high abundance and ubiquitous occurrence regardless of habitat type

107

(rainforest or pasture). Ants (Hymenoptera: Formicidae) were selected as a target taxon, and sorted to species. Where possible, ants were identified as described species using published taxonomic literature, otherwise they were assigned species codes. With assistance from Alan Andersen (CSIRO Sustainable Ecosystems), ant species were assigned to functional groups.

During the arthropod sampling, soil moisture content was measured by hand-collecting topsoil (up to 2-3 cm in depth) of approximately 50 cm3 from all of the experimental plots at five sites as well as their adjacent rainforest and pasture areas. Small amounts of topsoil were collected evenly over the plot area, or 2.5 x 2.5 m quadrats established within rainforest and pasture. Samples were collected between 15 April and 7 May 2004. The soil was kept in an airtight plastic bag, and was weighed before and after it was oven dried for 24 hours at 105 C˚. The ground temperature was recorded at 30 minute intervals for 5.5 days (14 to 20 April 2004) using temperature loggers (HOBO® Temperature Data Logger, Onset Computer Corporation, MA), deployed at the experimental plots and surrounding rainforest and pasture areas in one site only. The temperature sensor was placed on the ground surface (beneath the mulch, except in the un-mulched control plot), in the centre of each plot.

Data analysis Data were analysed in three different datasets: (i) arthropods sorted to Orders/Class (referred to as ‘coarse arthropods’ hereafter), (ii) ant species and (iii) ant functional groups. Each dataset was further divided into two separate subsets comprising data sampled using pitfall traps and litter extraction. Abundances of coarse arthropods were log transformed prior to analysis. Abundances of ant species were scored on a seven-point ordinal scale following Andersen et al. (2003): 1 = 1, 2 = 2-5, 3 = 6-20, 4 = 21-50, 5 = 51-100, 6 = 101-1000, 7 = >1000 individuals. Abundances within ant functional groups were expressed as proportions of all ants at each plot, and arcsine-transformed for analysis.

In addition to the data obtained from the present study, I also incorporated data from a preceding survey (Chapter 3), which provided baseline information on the distribution of arthropods in rainforest and pasture habitats of the study region. That survey was carried out in the same location and in a similar season, the previous year (8 January to 6 May 2003). Arthropods were collected from three sampling points at each of the 12 sites across

108

Maleny (N = 36), including the five sites used for the present study. Although the baseline arthropod sampling was carried out over a slightly larger area than in the present experimental study (each sampling point comprising a circular area of 3 m radius), sampling and sorting protocols were otherwise identical to those of the present study, so that direct comparison between the results of the baseline survey and the present study is possible.

Using the Indicator Value protocol (Dufrene & Legendre 1997), the baseline survey data were used to identify taxa that were characteristic of either rainforest or pasture habitats. These rainforest/pasture indicators were classed as either ‘specialist’ indicators (those which were found exclusively in their preferred habitat) or ‘increaser’ indicators (those which were found in both habitat types but were significantly more abundant in one). Other taxa that did not have significant habitat preferences were classed as ‘generalist’ (see Chapter 3 for more details).

Many individual habitat indicator taxa were, however, of limited usefulness due to their patchy distributions (Chapter 3). To develop a more robust indicator statistic, additional ‘composite rainforest/pasture indices’ were generated for each of the selected datasets (viz. coarse arthropods, ant species). To obtain the composite indices, the abundance values were summed across all taxa indicative of either rainforest (to calculate composite rainforest indices) or pasture (to calculate composite pasture indices) at each site, providing a single value which quantifies the extent to which a site resembled rainforest or pasture, in terms of its arthropod assemblage. Composite indices were calculated separately for coarse arthropods and ant species collected by either pitfall traps or litter extraction. Before calculating composite indices of coarse arthropods, abundance values of each of the component taxa were first individually range-standardised (site-specific abundance minus minimum abundance across all sites / maximum minus minimum) to remove the effects of large differences in taxon-specific abundance. This gave values between 0 and 1 for each taxon at each site. This range-standardisation procedure was not carried out for ant species, as most indicator species had a similar range in abundance score values (most ranging 0-4, with a maximum score of 6). Composite indices were calculated only if two or more of the component taxa/species were present.

Univariate analyses were carried out using two-factor crossed ANOVA with randomized complete block design to evaluate the response of variables (total abundances, taxon

109

richness, individual arthropod abundances, composite rainforest/pasture indices, soil moisture contents) to the experimental treatments. Factors tested were shading (0%, 50%, 90%), mulch depth (shallow, deep) and their interaction. Between-site variation (blocks) was also incorporated as a random factor. When the effect of shading was statistically significant, post-hoc paired-comparison, using a LSD test, was carried out. ANOVA was carried out on the abundance of an individual taxon only if that occurred in at least four of the total experimental plots used for the analyses (N = 30). To permit direct comparison with pasture and rainforest reference sites, I also present arthropod data from the baseline survey obtained at the same five sites (one of the three sampling points randomly selected at a site).

Nonparametric multivariate ANOVA (MANOVA) was carried out with PERMANOVA software (Anderson 2005) to test for responses of arthropod assemblages to the experimental treatments. This software executes MANOVA, using permutation methods, to calculate P values derived from pseudo F statistics of the distance measures (Anderson 2005). Factors being tested were the same as ANOVA. Between-site variation (blocks) was accounted for by including sites in the program as covariates, in a manner specified by M. Anderson (personal communication). When the effect of shading was statistically significant, post-hoc comparisons were carried out using pairwise permutation tests between shading levels.

6.3 Results

Overall abundances A total of 8839 arthropods were sampled from the experiment, but the majority were from pitfall traps (7162 individuals). Among 28 coarse arthropod taxa identified, ants were the most abundant with 3633 individuals (2897 from pitfall traps), followed by Coleoptera with 1846 and Araneae with 582. Fifty-two ant species were identified; however, 26 of them were rare species that occurred at less than four plots. Ants were assigned to eight functional groups, but most of them (95% of the total ant captures) belonged to Opportunists (1337 individuals), Cryptic species (991), Generalised Myrmicinae (953) or Dominant Dolichoderinae (155).

A significant effect of shading was found in total abundances of pitfall-trapped ants (ANOVA for the effect of shading and mulch depth: F = 4.90, P = 0.019; F = 1.28, P =

110

0.272 respectively); a post-hoc test showed that abundances were greater in 0% and 50% shading than in 90% shading. Neither the abundance or taxon richness of coarse arthropods nor the species richness of ants responded significantly to the shading or mulch depth treatments (results not shown).

Coarse arthropods Multivariate analyses using PERMANOVA showed statistically significant effects of shading on the compositions of pitfall-trapped coarse arthropod assemblages (Table 6.1). A post-hoc permutation test showed that the coarse taxonomic composition of arthropod assemblages under 0% shading differed significantly from those under both 50% and 90% shading. Shading and litter depth treatments did not have a significant influence on compositions of litter-extracted coarse arthropods (Table 6.1).

The composite index of rainforest ‘increasers’ based on pitfall-trapped coarse arthropods responded positively to the shading (Figure 6.1a, Table 6.1). Shading also had a significant negative effect on the composite index of pasture ‘increasers’: abundances of pasture-associated arthropods were lower in the plots under 90% shading than in plots without shading, with no significant interaction between shading and mulch depth (Figure 6.1c, Table 6.1). Composite habitat indices were not calculated for any habitat ‘specialists’ due to the very rare occurrences (N < 4) of the component taxa.

111

112

Table 6.1 Effects of shading and mulch depth on assemblage composition and composite rainforest and pasture indices of coarse arthropods. Between-site effects are also shown. ‘Difference’ shows the results of post-hoc permutation (for assemblage composition) or LSD (for composite indices) tests (% levels with different letters are significantly different, P < 0.05). Df for shading, mulch depth, interaction and site are 2, 1, 2, and 4 respectively. Composite habitat indices were not calculated for all ‘specialist’ indicators as they occurred at less than four plots, or were absent altogether. P value Shading Depth T† (S) (D) S x D Site Difference Assemblage composition Pitfall traps n/a 0.006 0.448 0.548 <0.001 0%(A) 50%(B) 90%(B) Litter extraction n/a 0.252 0.339 0.994 <0.001 - Composite rainforest index ‘Increaser' indicators (Pitfall traps) 9 0.045 0.669 0.993 0.003 0%(A) 50%(B) 90%(AB) ‘Increaser' indicators (Litter extraction) 14 0.337 0.320 0.483 0.110 - Composite pasture index ‘Increaser' indicators (Pitfall traps) 3 0.022 0.347 0.100 0.050 0%(B) 50%(AB) 90%(A) ‘Increaser' indicators (Litter extraction)‡ 1 - - - - - †Number of taxa used for that composite indicator. ‡Composite habitat indices were not calculated as only one taxon was found in these groups. Significant values (P < 0.05) are shown in bold.

a Composite RF index (‘Increasers’, Pitfall traps) b Composite RF index (‘Increasers’, Litter extraction) 5 6

4 4 3

2 Index level level Index Index level level Index 2 1

0 0 PastureP NoneNo 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% DM 90% 90% SM 90% 90% DM Rain- RF Pasture No 0% 0% 50% 50% 90% 90% Rain- Mulch S D S D S D forest P None 0% SM 0% DM 50% 50% 90% 90% RF Mulch S D SM S DM D SM S DM D forest

c Composite PA index (‘Increasers’, Pitfall traps) 2

1 Index level level Index

0 PastureP NoneNo 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% DM 90%90% SM 90% 90% DM Rain- RF Mulch S D S D S D forest

Figure 6.1 Effect of shading and mulch depth on composite rainforest (RF) and pasture (PA) indices of coarse arthropods (mean index level, SE). Values for rainforest and pasture reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars.

Among the pitfall-trapped coarse arthropods, two individual taxa were rainforest ‘specialists’ (Archaeognatha, Opilionida), as defined in Chapter 3; nine were rainforest ‘increasers’ (Blattodea, Coleoptera, Dermaptera, Diplopoda, Diplura, Heteroptera, Isopoda, Pseudoscorpionida, Psocoptera); and three were pasture ‘increasers’ (Araneae, Homoptera, Orthoptera). Among litter-extracted arthropods, there were 14 rainforest ‘increasers’ (Amphipoda, Blattodea, Chilopoda, Coleoptera, Dermaptera, Diplopoda, Diplura, Formicidae, Heteroptera, Isopoda, ‘other Hymenoptera’, Pauropoda, Pseudoscorpionida, Symphyla) and a single pasture ‘increaser’ (Orthoptera). Despite the large number of indicator taxa sampled, few of them showed statistically significant responses to the shading and litter depth treatments (Table 6.2). A number of ‘generalists’ showed significant responses to the shading treatments (Table 6.2), mostly showing elevated abundances in experimental plots at 0% and/or 50% shading (Figure 6.2), while abundances at 90% and in both rainforest and pasture were lower (although the latter two were sampled in a different year).

113

114

Table 6.2 Effects of shading and mulch depth on abundances of coarse arthropod taxa: results only for taxa where the main effects of ANOVA P ≤ 0.10. Between-site effects are also shown. ‘Difference’ shows the results of LSD tests (levels with different letters are significantly different; A smaller, B larger, P < 0.05). Df for shading, mulch depth, interaction and site are 2, 1, 2, and 4 respectively. P value Shading Depth Indicator category Freq† (S) (D) S x D Site Difference Pitfall traps Diplura Rainforest ‘increaser’ 21 0.060 0.088 0.936 0.001 - Homoptera Pasture ‘increaser’ 26 0.083 0.148 0.315 0.502 - Formicidae ‘Generalist’ 30 0.019 0.272 0.558 0.441 0%(B) 50%(B) 90%(A) Other Hymenoptera ‘Generalist’ 29 0.035 0.911 0.468 0.001 0%(A) 50%(B) 90%(A) Amphipoda ‘Generalist’ 24 0.098 0.760 0.496 0.011 - Pauropoda ‘Generalist’ 9 0.041 0.561 0.165 <0.001 0%(B) 50%(A) 90%(A) Thysanoptera ‘Generalist’ 17 0.009 0.677 0.915 0.112 0%(B) 50%(A) 90%(A)

Litter extraction Dermaptera RF ‘increaser’ 24 0.400 0.071 0.542 0.588 - Diplura RF ‘increaser’ 25 0.326 0.065 0.801 <0.001 - Thysanoptera ‘Generalist’ 4 0.025 0.877 0.976 0.561 0%(B) 50%(A) 90%(A) † Number of plots (N = 30) where that taxon was present. Significant values (P < 0.05) are shown in bold.

Pitfall traps

a Diplura b Homoptera (Rainforest ‘increaser’) (Pasture ‘increaser’) 14 10 12

8 10

6 8 6 4 Abundance Abundance 4 2 2

0 0 PastureP None No 0% 0% SM 0% 0% DM 50% 50% SM 50%50% DM 90%90% SM 90%90% DM Rain- RF PastureP NoneNo 0% 0% SM 0% 0% DM 50%50% SM 50%50% DM 90% 90% SM 90% 90% DM Rain- RF Mulch S D S D S D forest Mulch S D S D S D forest

c Formicidae d Other Hymenoptera (‘Generalist’) (‘Generalist’)

16 150

12

100 8 Abundance 50 Abundance 4

0 0 PastureP None No 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% DM 90%90% SM 90% 90% DM Rain- RF PastureP NoneNo 0% 0% SM 0% 0% DM 50% 50% SM 50%50% DM 90% 90% SM 90% 90% DM Rain- RF Mulch S D S D S D forest Mulch S D S D S D forest

e Amphipoda f Pauropoda (‘Generalist’) (‘Generalist’)

80 3

60 2

40 Abundance Abundance 1 20

0 0 P None 0% SM 0% DM 50% SM 50% DM 90% SM 90% DM RF Pasture No 0% 0% 50% 50% 90% 90% Rain- PastureP None No 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% 90% 90% SM 90% 90% Rain-RF Mulch S D S D S D forest Mulch S D S DMD S DM D forest g Thysanoptera (‘Generalist’)

4

3

2 Abundance 1

0 PastureP None No 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% DM 90% 90% SM 90% 90% DM Rain- RF Mulch S D S D S D forest

Figure 6.2 (2 pages) Effect of shading and mulch depth on abundances (mean, SE) of pitfall-trapped (a-g) and litter-extracted (h-j, next page) coarse arthropod taxa: results only for taxa where the main effects of ANOVA P ≤ 0.10 (see also Table 6.2). Values for rainforest and pasture reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars.

115

Litter extraction

h Dermaptera (Rainforest ‘increaser’) i Diplura (Rainforest ‘increaser’) 8 16

6 12

4 8 Abundance Abundance 2 4

0 0 PastureP NoneNo 0% 0% SM 0% 0% DM 50% 50% SM 50%50% DM 90% 90% SM 90% 90% DM Rain- RF PastureP None No 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% DM 90% 90% SM 90% 90% DM Rain- RF Mulch S D S D S D forest Mulch S D S D S D forest

j Thysanoptera (‘Generalist’)

12 10 8 6

Abundance 4 2 0 PastureP NoneNo 0% 0% SM 0% 0% DM 50% 50%50% 90%90% 90%90% Rain-RF Mulch S D SM S DM D SMS DM D forest

Figure 6.2 (cont’d)

Ant species Shading had a significant effect on species composition of pitfall-trapped ant assemblages (Table 6.3). The response was similar to that observed for coarse arthropods: assemblages of pitfall-trapped ant species in plots under 0% shading were different from those under 90% shading, and assemblages under 50% shading were intermediate (Table 6.3).

Within pitfall-trapped ants, the composite index value for rainforest ‘specialists’ was greater in plots with shallow than deep mulch (Table 6.3, Figure 6.3a). No experimental treatments significantly influenced the composite index of rainforest ‘increasers’ (Table 6.3, Figure 6.3b). Levels of the composite index of pasture ‘specialists’ were significantly lower in plots with 90% shading compared to those with 0% shading (Table 6.3, Figure 6.3c). Composite habitat indices were not calculated for litter-extracted ants, as only one component species was identified for each group of habitat indicator.

116

Table 6.3 Effects of shading and mulch depth on assemblage composition and composite rainforest and pasture indices of ant species. Between-site effects are also shown. ‘Difference’ shows the results of post-hoc permutation (for assemblage composition) or LSD (for composite indices) tests (% levels with different letters are significantly different, P < 0.05). Df for shading, mulch depth, interaction and site are 2, 1, 2, and 4 respectively. Composite indices were not calculated for all litter-extracted ants, as less than 2 component species (T† < 2) were found from each index. P value Shading Depth T† (S) (D) S x D Site Difference Assemblage composition Pitfall traps n/a 0.025 0.851 0.510 <0.001 0%(A) 50%(AB) 90%(B) Litter extraction n/a 0.387 0.676 0.276 0.025 - Composite rainforest index ‘Specialist' indicators (Pitfall traps) 5 0.178 0.049 0.335 0.002 Deep(A) Shallow(B) ‘Increaser' indicators (Pitfall traps) 2 0.858 0.441 0.858 0.244 - Composite pasture index ‘Specialist' indicators (Pitfall traps) 3 0.048 0.117 0.984 0.009 0%(B) 50%(AB) 90%(A) ‘Increaser' indicators (Pitfall traps) ‡ 1 - - - - - †Number of taxa used for that composite indicator. ‡Composite habitat indices were not calculated as only one taxon was found in these groups. Significant values (P < 0.05) are shown in bold.

117

a Composite RF index (‘Specialists’, Pitfall traps) b Composite RF index (‘Increasers’, Pitfall traps) 2 6 1.5

4 1 Index level Index Index level level Index 2 0.5

0 0 PastureP None No 0% 0% SM 0% 0% DM 50%50% SM 50% 50% DM 90%90% SM 90% 90% DM Rain- RF PastureP NoneNo 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% DM 90% 90% SM 90% 90% DM Rain- RF Mulch S D S D S D forest Mulch S D S D S D forest c Composite PA index (‘Specialists’, Pitfall traps) 10

8

6

4 Index level Index 2

0 PastureP NoneNo 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% DM 90% 90% SM 90% 90% DM Rain- RF Mulch S D S D S D forest

Figure 6.3 Effect of shading and mulch depth on composite rainforest (RF) and pasture (PA) indices of ant species (mean index level, SE). Values for rainforest and pasture reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars.

Among pitfall-trapped ants, there were six rainforest ‘specialists’, as defined in Chapter 3 (Anonychomyrma QM3, Leptomyrmex erythrocephalus rufithorax, Monomorium tambourinense, Pheidole QM1, Pheidole QM2, Pheidole sp.2), two rainforest ‘increasers’ (Notoncus capitatus, Rhytidoponera chalybaea), three pasture ‘specialists’ (Rhytidoponera metallica, Pheidole QM3, Cardiocondyla nuda), and one pasture ‘increaser’ (Carebara QM1). Among litter-extracted ants, a single species of each group was found: rainforest ‘specialist’ (Carebara QM2), rainforest ‘increaser’ (Hypoponera sp.1), pasture ‘specialist’ (Pheidole QM3), and pasture ‘increaser’ (Carebara QM1). As the occurrences of most indicators were patchy, statistical analyses were carried out on only eight indicators. Within rainforest indicators, only Hypoponera sp.1 (litter-extracted rainforest ‘increaser’) significantly responded to the experimental treatments (Table 6.4), progressively increasing in abundance with increased shading (Figure 6.4f). Three ‘generalist’ species (Paratrechina QM1, Pheidole QM8, Solenopsis QM1) had been very rare or absent in both rainforest and pasture sites sampled the previous year, but were abundant in experimental plots, especially in less shaded plots during the year of this study (Table 6.4, Figure 6.4c-e).

118

Table 6.4 Effects of shading and mulch depth on abundance scales of ant species: results only for species where the main effects of ANOVA P ≤0.10. Between-site effects are also shown. ‘Difference’ shows the results of LSD tests (levels with different letters are significantly different; A smaller, B larger, P < 0.05). Df for shading, mulch depth, interaction and site are 2, 1, 2, and 4 respectively. P value Shading Depth Indicator category Freq† (S) (D) S x D Site Difference Pitfall traps Anonychomyrma QM3 Rainforest ‘specialist’ 8 0.328 0.060 0.094 0.043 - Rhytidoponera metallica Pasture ‘specialist’ 18 0.174 0.060 0.539 0.003 - Paratrechina QM1 ‘Generalist’ 13 0.051 0.838 0.026 0.051 - Pheidole QM8 (mjobergi grp.) ‘Generalist’ 10 0.022 0.652 0.288 <0.001 0%(B) 50%(A) 90%(A) Solenopsis QM1 ‘Generalist’ 28 0.051 0.503 0.857 <0.001 0%(AB) 50%(B) 90%(A)

Litter extraction Hypoponera sp.1 Rainforest ‘increaser’ 8 0.044 0.210 0.871 0.099 0%(A) 50%(AB) 90%(B) † Number of plots (N = 30) where that species was present. Significant values (P < 0.05) are shown in bold.

119

Pitfall traps a Anonychomyrma QM3 b Rhytidoponera metallica (Rainforest ‘specialist’) 2 (Pasture ‘specialist’)

3

1 2

1 Abundance scale scale Abundance Abundance scale scale Abundance

0 0 PastureP None No 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% DM 90% 90% SM 90% 90% DM Rain- RF PastureP None No 0% 0% SM 0%0% DM 50% 50% SM 50% 90% 90% SM 90% 90% Rain-RF Mulch S D S D S D forest Mulch S D S DM D S DM D forest c Paratrechina QM1 d Pheidole QM8 (‘Generalist’) (‘Generalist’) 4 3 3 2 2

1

Abundance scale scale Abundance 1 Abundance scale scale Abundance

0 0 PastureP None No 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% 90% 90% SM 90% 90% Rain-RF PastureP None No 0% 0% SM 0% 0% DM 50% 50% SM 50%50% DM 90%90% SM 90%90% DM Rain- RF Mulch S D S DM D S DM D forest Mulch S D S D S D forest

e Solenopsis QM1 (‘Generalist’) 4

3

2

1 Abundance scale

0 PastureP NoneNo 0% 0% SM 0% 0% DM 50% 50% SM 50% 50% 90% 90% SM 90% 90% Rain-RF Mulch S D S DM D S DM D forest

Litter extraction f Hypoponera sp.1 (Rainforest ‘increaser’)

3

2

1 Abundance scale

0 PastureP None No 0% 0% SM 0%0% DM 50% 50% 50%50% 90%90% 90%90% Rain-RF Mulch S D SMS DDM SMS DM D forest

Figure 6.4 Effect of shading and mulch depth on abundance scales of ant species: results only for species where the main effects of ANOVA P ≤ 0.10 (see also Table 6.4). Values for rainforest and pasture reference habitats (sampled the previous year) are also shown. The plots included in the statistical analysis were represented by closed bars.

120

None of the ant functional groups responded significantly to the experimental treatments. The strongest response within ant functional groups was that pitfall-trapped Cryptic species showed a non-significant trend to increase in relative abundance with increased mulch depth (ANOVA for the effect of shading and mulch depth: P = 0.226, P = 0.083 respectively; interaction P = 0.520).

Soil moisture content and temperature Mean soil moisture content was higher in pasture (at least 100 m away from the forest edge) than in rainforest (Figure 6.5). Compared with the un-mulched control plots at the forest edge, a relatively high soil moisture content was maintained by all of the experimental plots regardless of differences in shading and mulch depth; however, none of the plots were as moist as the pasture or rainforest reference sites (Figure 6.5). No significant effects of shading or mulch depth were found on soil moisture content across the experimental treatments (two-factor ANOVA for the effect of shading and mulch depth: P = 0.885, P = 0.158 respectively; interaction P = 0.670), whereas a single factor ANOVA found significant differences among pasture, no mulch, rainforest, and the combined experimental treatments (P < 0.001), with pairwise tests showing that all were different.

60

50

40

30

20

Soil moisture (%) 10

0 PasturePasture No 0m0%ShallowWood 0% 0% 0m50%ShallowWood 50% 50% 0m90%ShallowWood 90% 90% Rain-RF Mulch S D S D S D forest

Figure 6.5 Average soil moisture contents across the experimental treatments and unshaded, un-mulched controls. Values for rainforest and pasture reference habitats are also shown. All samples were taken in 2004. The plots included in the two-factor ANOVA were represented by closed bars.

121

Temperatures recorded in rainforest were lower on average than those in pasture, and the coefficient of variation of half-hourly temperatures over five days in rainforest was about half that of pasture (Figure 6.6a). The provision of mulch in the experimental plots suppressed extreme temperature fluctuations (compare the control plot in Figure 6.6a with Figure 6.6b). The presence of shading further reduced average temperatures and their coefficients of variation (Figure 6.6b).

Mean CV Figure 6 Control 20.84 23.92 a Pasture 20.02 6.23 35 Rainforest 18.28 3.50 ) ˚ 30

25

Temperature (C Temperature 20

15 0 20 40 60 80 100 120 140 Mean CV 0% Shallow 23.21 7.66 b 0% Deep 22.15 8.43 35 50% Shallow 20.01 4.63 50% Deep 21.18 4.52 90% Shallow 19.52 3.52 90% Deep ˚ ) 30 20.02 3.64

25

Temperature (C Temperature 20

15 0 20406080100120140 Time lapsed (hour)

Figure 6.6 Temperature fluctuations over five days recorded at: (a) un-mulched, unshaded control plots, rainforest and pasture reference habitats, and (b) plots with different shading and mulch depths, with mean values and coefficient of variations (CV). Temperature was recorded only from one site (E3, D3) during April 2004. Horizontal lines are drawn at 25 C˚.

122

6.4 Discussion

Effects of the experimental context and sampling methods To date, most restoration studies have employed a post-hoc empirical approach to investigate the effects of factors considered important for the development of colonising fauna (Majer 1989a; Michener 1997, see also Section 1.3.3 ). The ecological effects of the factors under investigation are, however, often difficult to elucidate, due to the presence of extraneous factors (Block et al. 2001; Catterall et al. 2004). This problem is further exacerbated by the fact that restoration projects are generally carried out on a site-specific basis with no spatial replication, limiting inference from the data (Block et al. 2001). In contrast, the present study provided an opportunity to systematically test the factors of interest since the experimental approach allowed for the construction of replicated units in which focal factors (i.e. shading and litter depth) were manipulated, while controlling for extraneous factors.

There were, however, a number of constraints associated with the experimental design, which potentially limited the successful colonisation of constructed plots by rainforest-dependent arthropods. First, experimental plots lacked some of the habitat components of reforested sites, namely live plants that supply freshly shed foliage, and woody debris, both of which may potentially be important for the colonisation and persistence of rainforest-dependent arthropods (Majer et al. 1984; Andrew et al. 2000; Grove 2000). Second, the mulch used in the experiment had been sterilised with steam, which may have killed the arthropods’ potential food resources (e.g. prey micro-invertebrates, bacteria, fungi). Third, the spatial and temporal scale of the experiment may have been insufficient to allow successful colonisation to occur, although the plots’ location adjacent to the forest edge would have maximised the probability of rainforest-associated taxa moving into the plots (see also Chapter 4). Despite any limitations, a diverse array of arthropods, including rainforest-dependent taxa, did colonise the experimental plots and their assemblage composition responded differentially to the varying experimental treatments.

Significant responses to the experimental treatments primarily involved pitfall-trapped arthropods, which would have represented the active epigaeic arthropod community at sample sites (Majer 1997; Bestelmeyer et al. 2000). The few responses of litter-extracted arthropods to treatments may, in part, be explained by the sampling intensity of litter

123

extraction, which yielded a lower number of arthropods (1677 individuals) than did pitfall trapping (7162 individuals). Alternatively, litter-extracted arthropods may have been slower to colonise and establish in experimental plots. The low rate of colonisation was particularly apparent in rainforest species of litter-extracted ants (e.g. Mayriella abstinens complex, Strumigenys harpyia, Discothyrea, Lordomyrma spp., see Chapter 3, Appendix 3b), which were either very low in abundance or absent from the experimental plots.

Responses of arthropods to shading and litter depths Previous studies of the development of soil and litter fauna in revegetated sites have often reported an initial colonisation by species presumably tolerant of harsh environmental conditions (e.g. Fox & Fox 1982; Andersen 1993; van Aarde et al. 1996a; Dunger et al. 2001). As the restored habitat developed, these species were gradually or abruptly replaced by other species characteristic of undisturbed habitats. In the context of rainforest restoration, the results of the present study suggest that successional patterns of this type could occur as a function of increased shading alone. Shading can occur because of the development of canopy closure – one of the attributes most strongly associated with the development of restored rainforests (Kanowski et al. 2003). The effects of shading revealed by the present study are consistent with the findings of a growing number of other ecological studies suggesting the importance of structural attributes, including canopy cover, for the organization of soil and litter arthropod assemblages (e.g. Holmes et al. 1993; Watts & Gibbs 2002; Proctor et al. 2003; Lassau & Hochuli 2004; Grimbacher et al. 2007). Exceptions to this pattern may include some groups of arthropods that are linked strongly with biological traits of live plant species (e.g. herbivores, which show strong host-plant affinities, Hunter & Price 1992; Wardle 2006).

The observed patterns of arthropod colonisation may be due at least in part to improved temperature regimes in plots under shading. Temperature tolerance has been considered important in influencing the distribution of ground dwelling arthropods (Pearson & Lederhouse 1987; Parmenter et al. 1989; Addo-Bediako et al. 2000), and a number of restoration studies indeed suggested that arthropods characteristics of rainforests may prefer cooler microclimatic conditions (King et al. 1998; Grimbacher et al. 2006). The present study showed that lower average temperatures and reduced temperature fluctuations in the more heavily-shaded treatments were associated with an increased abundance of rainforest-associated arthropods and a reduced abundance of

124

pasture-associated arthropods. In contrast, soil moisture content did not appear to explain the colonisation patterns of arthropod assemblages (as it did not vary among the experimental plots), even though small arthropods could arguably be sensitive to reduced moisture levels (Levings & Windsor 1984; Shure & Phillips 1991). As the sampling was conducted during late summer season when conditions are relatively hot and wet, mitigation of temperature regimes may have been more important in shaded plots. Had I conducted sampling in winter (typically cool, dry), the patterns and role of soil moisture content may have been different. However, arthropod activity and abundance may also be low at all sites in winter.

In addition to the effects of microclimatic conditions, colonisation patterns, particularly those of pasture-associated arthropods, may have been affected by the growth of herbaceous vegetation. For example, high levels of colonisation by pasture-associated coarse arthropod taxa were observed in unshaded plots, and in plots under 50% shading with shallow mulch (Figure 6.1c). In all these unshaded or lightly shaded plots there was visible regrowth of herbaceous pasture plants (see also Plate 7), which potentially provided of food and habitat for pasture-associated herbivorous arthropods, such as Homoptera and Orthoptera. With regard to pasture ant species, colonisation patterns may have also been influenced by the availabilities of some food resources (e.g. seeds from pasture, honeydew from Homoptera) associated with the growth of the pasture plants.

Among previous restoration studies of soil and litter arthropods in actively revegetated sites (Table 1.5), at least 25 investigated the effects of differing levels of canopy cover and/or litter amount on the development of soil and litter arthropod assemblages. While these studies frequently found shading to be significant or important, many did not find an impact due to the amount of litter. Large proportion of the studies (including those conducted in the context of rainforest restoration, Jansen 1997; King et al. 1998, see Appendix 1) found no apparent responses in the rate of arthropod colonisation to different amount of forest litter. This study’s results were consistent with this pattern, suggesting that the amount of litter may not be a strong determinant of arthropod organisation in the context of rainforest restoration. Although rainforest-like arthropod assemblages may require the presence of at least a thin layer of litter similar to that typically found on the rainforest floor (ca. 3 cm, see King et al. 1998; Nakamura et al. 2003), this study’s results demonstrated that addition of extra mulch does not benefit their colonisation.

125

The only variable that responded significantly to changes in mulch depth was composite rainforest index based on pitfall-trapped ants, with levels higher in plots with shallow mulch. Another aspect of this same field experiment (Chapter 5) that used a different variety of mulch (primarily monocultural hay mulch) at two different depths (shallow, 3-5 cm; deep, 10-15 cm) also found that the shallow mulch treatment was favoured by ant assemblages characteristic of rainforest. The results of these studies suggest that provided that other habitat conditions (e.g. temperature regimes) are suitable, shallow mulch (3-5 cm) may facilitate colonisation by rainforest-like ant assemblages more effectively than deep mulch (10-15 cm).

Management implications Irrespective of the goals of rainforest restoration, the re-establishment of soil and litter dwelling arthropods is undoubtedly important due to their considerable abundance, diversity and roles in ecosystem functioning (Wilson 1987). In order to maximise ’rainforest biodiversity values’ (sensu Catterall et al. 2004), restored habitat patches need to facilitate colonisation by fauna characteristic of rainforests, while inhibiting the (re-)invasion by arthropods characteristic of the matrix habitat (i.e. pasture). This study’s results suggest that rainforest restoration using lower density plantings, such as timber plantations, may facilitate colonisation by rainforest arthropods even though canopy cover may not be developed as rapidly, or to the same extent, as in ecologically-designed restoration plantings or rainforest. However, the establishment of a fully closed canopy (90%) appeared to most effectively inhibit invasion by arthropods characteristic of matrix habitat (i.e. pasture). There was no evidence that the addition of extra mulch compensated for reduced canopy cover. Indeed, it appeared that a deep mulch layer did not necessarily create more hospitable conditions for arthropods characteristic of rainforest, than a shallow mulch layer more characteristic of rainforest.

126

7 EFFECTS OF HERBICIDE APPLICATION

7.1 Introduction The use of herbicide for removal of plant cover, including control of weeds, is an integral part of management strategies in many instances of terrestrial restoration (Goosem & Tucker 1995; Kooyman 1996; Lyons et al. 2000; Woodford 2000; Gratton & Denno 2005). Glyphosate is a systemic broad-spectrum herbicide which has been used extensively in a variety of restoration programs, as it has a reputation not only for cost effectiveness and strong efficacy, but also low toxicity to non-target organisms (Grossbard & Atkinson 1985). Giesy et al. (2000) carried out an extensive review of the toxicity of glyphosate herbicides on selected species of terrestrial and aquatic organisms. With regard to terrestrial organisms, they concluded that, although glyphosate and commercially available glyphosate formulations were slightly to moderately toxic to a number of species under laboratory conditions, the herbicides were effectively non-toxic if used at the recommended concentrations in the field. Their claim is further supported by the fact that, when applied to soil, glyphosate is inactivated by adsorption to clay minerals, and both bound and free glyphosate is degraded by micro-organisms (Sprankle et al. 1975b; Torstensson 1985; Glass 1987).

Despite broad consensus on the low toxicity of glyphosate herbicides, further investigation is warranted as their toxicity under field conditions may be much greater than predictions based on a limited number of mostly laboratory-based toxicological studies of recommended application rates. Cornish and Burgin (2005) argued that the actual amount of herbicide applied may be highly variable and substantially exceed the maximum recommended dose when operators do not abide strictly by regulatory requirements – often the case in small scale restoration activities. They further argued that a considerable proportion of the herbicide may not be absorbed by plants and inevitably will reach soil where there is incomplete foliage cover, or where rainfall occurs after application. Field experiments that simulate the realistic application of glyphosate herbicide are, therefore, needed to test toxicity to non-target terrestrial organisms in the context of forest restoration.

Amongst the array of non-target organisms, soil and litter dwelling arthropods are arguably one of the most susceptible to the potential impacts of the herbicide as they are

127

strongly associated with the substrate where herbicide residues may accumulate (van Straalen 1998). These arthropods include herbivores and detritivores that take in the herbicide through direct consumption of contaminated plant or litter (e.g. Amphipoda, Diplopoda, Homoptera, Isopoda, Orthoptera; Ruppert et al. 2004). Ants may also be informative subjects as a number of studies have shown ants to be bioindicators of contamination by various types of pollutants in the field (Madden & Fox 1997; Hoffmann et al. 2000; but see Folgarait 1998). If they are susceptible to herbicide contamination, then ants would contribute an effective bioassay given that the biology, ecology and taxonomy of ants are relatively well known (Holldobler & Wilson 1990; Agosti et al. 2000).

A limited number of studies have been carried out to test experimentally for the ecotoxicological effects of glyphosate herbicides under field conditions (Brust 1990; Haughton et al. 1999b; Haughton et al. 1999a; Haughton et al. 2001a). These studies found that although glyphosate application effectively had no to slight toxicity to selected groups of epigaeic arthropods under laboratory conditions, it had significant impacts on these arthropods under field conditions. The results of these field experiments, however, were dominated by the indirect effects of vegetation loss. Glyphosate spray leads to the death of plants (the aim of herbicide application), which then affects arthropod assemblages indirectly through environmental changes, such as moisture availability, temperature regimes, floral diversity and food resource availability (House et al. 1987; Lindsay & French 2004). The direct toxicity of glyphosate herbicides to soil and litter arthropods under field conditions has not been assessed explicitly.

This chapter has investigated the ecotoxicological effects of a commercially available glyphosate formulation (Roundup® Biactive™) on assemblages of soil and litter dwelling macro-arthropods, with particular attention to ant species, in subtropical rainforests on the Maleny plateau of eastern Australia. The experimental design incorporated conditions similar to those used in rainforest restoration activities within the region, and differed in two ways from previous field experiments. First, the herbicide was applied at the rate of 20 L product/ha (product with 360 g glyphosate/L), which is considerably greater than the maximum recommended application rate of 9 L product/ha for hard-to-kill weeds. This was done to simulate worst-case scenarios that may occur in the actual activities of forest restoration (Cornish & Burgin 2005). Second, my field experiment was carried out under dense canopy cover with sparse understorey vegetation, 128

so that potentially large impacts caused by the loss of existing vegetation were eliminated.

7.2 Methods

Study area The study was undertaken on the Maleny plateau, in the Sunshine Coast hinterland of eastern Australia (26° 40’- 50’ S, 152° 45’- 53’ E, elevation 350 to 530 m). Mean daily maximum and minimum temperatures in mid summer (January) are 28.9 and 18.8 ° C respectively, and 19.5 and 7.1 ° C in mid winter (July). Average annual rainfall in the region is 1973 mm, with most falling between December and April (climate data averaged over at least 50 years up to 2004 obtained from the Australian Bureau of Meteorology). The total precipitation during the study period (1146 mm between December 2003 and May 2004) was below the long term average (1410 mm).

Following European settlement of the Maleny plateau, extensive areas of subtropical rainforest were cleared, and large parts are now farms or residential areas. Recently, however, community and government initiatives in this region have lead to the extensive efforts to restore rainforests in formerly cleared land (Catterall & Harrison 2006).

Five study sites (E1-E5) were located within separate rainforest remnants dispersed across a study region of around 170 km2; all within 13 km of the township of Maleny. The minimum distance between rainforest remnants was 1 km; most sites were > 2 km apart (Figure 2.1). The size of the smallest rainforest remnant was 1.15 ha, but the other four rainforests were well over 10 ha. Rainforest remnants were either old regrowth (age of ca.100 years) or remnants that had been selectively logged until recently (Appendix 2). All rainforest remnants had well established canopy cover of over 90%. Structural types (Adam 1992) of the rainforests were either complex notophyll vine forest, or (for one site located on the escarpment of the plateau) notophyll feather palm vine forest. Four sites were on basaltic soil; a combination of basaltic soil and metamorphic rocks was present at the other site located on escarpment of the plateau. All rainforest remnants were abutted by pasture dominated by kikuyu grass (Pennisetum clandestinum).

129

Experimental design At each experimental site, I established paired plots, each consisting of a 3 x 3 m quadrat located at least 30 m from the remnant edge. One plot of each pair received herbicide spray and the other was set as a control which did not receive herbicide. The treatment and control plots of each pair were separated by at least 30 m. When a rainforest remnant was located on a slope, control plots were located upslope to avoid possible cross-contamination. None of the study sites had a known history of herbicide application.

The concentrated solution of the herbicide Roundup® Biactive™ (360 g/L glyphosate with undisclosed surfactant, Monsanto 2000), was diluted with potable water to 7.2 g/L glyphosate. In each treatment plot, approximately 900 ml of the diluted herbicide was applied with a pressure sprayer (Yates® Maxi 6 Litre Garden Sprayer), ensuring that the entire forest floor within a plot received herbicide application. Care was taken to avoid direct contact of the herbicide with understorey vegetation and seedlings which occurred sparsely within each plot. Herbicide was applied between 29 December 2003 and 23 January 2004, ensuring that there was no rainfall at least 12 hours after application.

Sampling methodology Three sampling events occurred at each plot: before the herbicide application (5-13 days before application, 23 December 2003-11 January 2004); 2-4 days after application (31 December 2003-27 January 2004); and 86-103 days after application (9 April-2 May 2004). These sampling events are hereafter referred to as ‘before application’, ‘three days after application’ and ‘three months after application’.

At each plot arthropods were collected using litter extraction by Tullgren funnels. This sampling method primarily collects less mobile arthropods and was, therefore, expected to best detect the effects of the herbicide within the limited area (3 x 3 m quadrat) of each experimental plot. One litre of litter and surface soil (to a depth of 1 to 2 cm) was collected in small amounts evenly over the entire plot area. The same proportion of surface soil relative to litter was collected (approx. 20% surface soil and 80% litter by volume). Each soil and litter sample was placed in a fabric bag and kept in an insulated box. Samples were placed in Tullgren funnels within 12 hours of sampling, and extracted for 4.5 days using 40 W clear light bulbs.

130

Identification of macro-arthropods was to Order except a) Hymenoptera which were split into Formicidae and others and b) myriapods to Class. Macro-arthropods are here defined as all Arthropoda except the ubiquitous Acari and Collembola. Ants (Hymenoptera: Formicidae) were selected as a target group, and sorted to species. Where possible, ants were identified as described species using published taxonomic literature, otherwise they were assigned species codes.

Data analysis Data were analysed in two different datasets: macro-arthropods sorted to Orders/Class (referred to as ‘coarse arthropods’ hereafter) and ant species. Abundances of both coarse arthropods and ant species were log transformed before analysis. For multivariate analyses of assemblage composition, both log-transformed abundances and presence/absence data were used.

Univariate analyses were carried out using repeated-measures ANOVA to evaluate the responses of variables (total abundances, taxon/species richness, individual arthropod abundances) to the experimental treatments. The five rainforest sites were the subjects. Within-subject factors tested included control/treated (control versus treated plots) and time after application (before, three days after, three months after application). Degrees of freedom for all F tests were corrected by Huynh-Feldt Epsilon values, whether or not each variable tested violated the assumption of sphericity, as recommended by Quinn and Keough (2002). ANOVA was carried out on the abundance of individual taxon/species only if that taxon occurred in at least four of the total experimental plots used for the analyses (N = 30).

Multivariate analyses were carried out using non-metric multi-scaling ordination (NMDS) with PRIMER v.5 software (Clarke 1993). All NMDS ordinations were performed on Bray-Curtis distance matrices, with 10 random restarts. Analysis of Similarity (ANOSIM) was used to test for differences in the composition of arthropod assemblages between treated and control plots. To compare the extent of changes in composition between control and treated plots after the herbicide application, the following procedure was used. First, I calculated Bray-Curtis distance values that measured dissimilarities of arthropod assemblage compositions before versus after (three days or three months after) application. The differences in the Bray-Curtis distance values were then compared between control and treated plots using randomization-based

131

paired-t-tests with 5000 permutations. Two separate tests (before versus three days after application, before versus three months after application) were carried out to assess the short- and long-term responses of assemblage compositions.

It should be noted that family-wise Type I error was not controlled and the significant P threshold was set at 0.05, because the present study is designed to screen for taxa that responded to the herbicide treatment (Roback & Askins 2005).

7.3 Results A total of 4003 arthropods were sampled. Among 24 coarse arthropod taxa identified, ants were the most abundant with 877 individuals, followed by Coleoptera with 832 and Isopoda with 590. Thirty-five ant species were identified; Hypoponera sp.1 was the most abundant (439 individuals), making up 52% of the total ant catches. Although many coarse arthropod taxa occurred abundantly at more than half of the total number of samples (N = 30), many ant species (22 species) were rare, occurring in less than four samples.

NMDS ordinations (Figure 7.1) and ANOSIM analyses of differences in assemblages (Table 7.1) of both coarse arthropods and ant species showed no overall separations between control and herbicide-treated plots regardless of the sampling time. Randomization-based paired-t-tests also showed no significant differences in the dissimilarity values of arthropod assemblage compositions between control and treated plots, suggesting that arthropod assemblage composition did not change significantly in response to immediate (three days after application) or delayed (three months after application) effects of the herbicide application (Table 7.2).

Effects of the herbicide application on taxon/species richness or abundances of individual arthropod taxa were indicated in an interaction of the two factors (control/treatment x time after application) in the results of the repeated-measures ANOVAs; however, interaction effects were not significant on any of the variables tested (Table 7.3). Abundances of Diptera declined three days after application (Figure 7.2f); however, its interaction effect was non-significant (Table 7.3). The variations in abundances of other individual taxa/species were consistent over time for both control and treated plots (Figure 7.2).

132

a Coarse arthropods b Ant species

Legend

Control, Before Control, 3 days after application Control, 3 months after application Treated, Before Treated, 3 days after application Treated, 3 months after application

Figure 7.1 NMDS ordinations based on log-transformed abundances of (a) coarse arthropods and (b) ant species before and after the herbicide application (three days and three months after application) for the control and herbicide-treated plots. NMDS ordinations based on presence/absence data are not shown as they produced similar patterns to those based on log-transformed abundance data.

Table 7.1 ANOSIM global R values comparing arthropod assemblage compositions of control versus herbicide-treated plots. Arthropod compositions were compared using all samples, samples collected before application, three days after application and three months after application. Both log-transformed abundance and presence/absence data were used for coarse arthropods and ant species.

Coarse arthropods Ant species

Global R P Global R P All Log(x+1) -0.061 0.954 0.004 0.435 Presence/absence -0.058 0.910 0.007 0.404

Before application Log(x+1) -0.144 0.897 0.004 0.444 Presence/absence 0.006 0.389 0.028 0.397

Three days after application Log(x+1) -0.220 0.984 -0.156 0.881 Presence/absence -0.156 0.897 -0.108 0.698

Three months after application Log(x+1) -0.112 0.786 0.036 0.349 Presence/absence -0.104 0.754 0.110 0.198

133

134

Table 7.2 Mean and standard errors of Bray-Curtis distance values that measured dissimilarities of arthropod assemblage compositions before versus after the herbicide application (three days or three months after application) at each site. Differences in the distance values between control and herbicide-treated plots were tested using randomization-based paired-t-test. Both log-transformed and presence/absence data were used for coarse arthropods and ant species.

Dissimilarities of assemblage Dissimilarities of assemblage compositions compositions before versus three days before versus three months after after application application

Control Treated Control Treated Mean SE Mean SE P Mean SE Mean SE P Coarse arthropods Log(x+1) 20.8 3.13 20.6 2.46 0.944 23.2 2.35 24.9 2.39 0.377 Presence/absence 20.6 1.92 17.3 3.18 0.496 18.1 3.30 19.2 4.90 0.810

Ant species Log(x+1) 46.7 3.54 55.5 4.42 0.190 51.5 8.15 64.1 7.65 0.316 Presence/absence 52.3 7.36 56.1 5.42 0.632 50.7 4.15 67.0 3.86 0.128

Table 7.3 (2 pages) Effects of control/treated (control versus treated) and time (before, three days, three months after application) on abundances and taxon richness of (a) coarse arthropods and (b) ant species (next page): results for variables that occurred in at least four samples (N = 30). The effect of the herbicide application was indicated by the interaction of the two factors. Significant values (P < 0.05) are shown bold.

a. Coarse arthropods

Control/treatment Time Interaction Freq† df F P df F P df F P Total abundance - 1 1.10 0.354 1.8 1.29 0.327 1.2 0.41 0.586 Taxon richness - 1 0.82 0.417 1.8 1.20 0.349 1.5 0.86 0.435

Amphipoda 21 1 0.23 0.654 2.0 2.64 0.132 1.7 0.80 0.468 Araneae 29 1 0.12 0.746 2.0 4.40 0.051 1.8 0.27 0.748 Blattodea 11 1 1.24 0.328 1.3 1.02 0.385 1.7 0.37 0.665 Chilopoda 27 1 5.02 0.089 2.0 0.34 0.724 1.4 0.19 0.759 Coleoptera 30 1 0.01 0.926 2.0 1.63 0.254 2.0 0.73 0.510 Dermaptera 13 1 5.32 0.082 1.7 3.11 0.114 2.0 0.69 0.528 Diplopoda 30 1 0 0.965 1.6 1.39 0.305 1.2 1.34 0.315 Diplura 18 1 0.12 0.749 1.6 0.34 0.679 1.2 0.00 0.974 Diptera 12 1 8.52 0.043 1.7 0.07 0.904 2.0 2.86 0.115 Formicidae 30 1 1.13 0.348 1.6 0.30 0.705 1.2 0.79 0.444 Heteroptera 30 1 2.83 0.168 2.0 0.76 0.498 1.2 0.33 0.628 Homoptera 17 1 3.01 0.158 1.5 1.43 0.298 2.0 1.68 0.246 Isopoda 30 1 1.29 0.319 1.8 1.21 0.347 2.0 0.07 0.937 Opilionida 4 1 1.00 0.374 1.9 1.56 0.271 1.0 1.00 0.374 Other Hymenoptera 14 1 2.19 0.213 1.3 1.67 0.262 2.0 1.55 0.270 Pauropoda 27 1 0.41 0.556 2.0 6.23 0.023 2.0 0.29 0.755 Protura 7 1 1.02 0.370 1.4 1.13 0.361 1.2 1.18 0.345 Pseudoscorpionida 26 1 0.44 0.542 1.2 0.43 0.581 2.0 0.21 0.812 Psocoptera 5 1 1.00 0.767 2.0 3.90 0.066 1.6 0.60 0.541 Symphyla 13 1 0.72 0.444 2.0 2.40 0.153 1.4 1.38 0.308 Thysanoptera 9 1 0.24 0.652 1.7 0.21 0.787 2.0 0.17 0.851 135

136

Table 7.3 (cont’d)

b. ant species

Control/treatment Time Interaction Freq† df F P df F P df F P Total abundance - 1 1.11 0.352 1.5 0.16 0.802 1.2 0.71 0.465 Species richness - 1 3.47 0.136 1.5 0.23 0.736 1.4 0.86 0.432

Carebara QM1 (C) 4 1 0.57 0.491 1.0 1.54 0.282 1.1 0.45 0.560 Discothyrea QM1 (C) 13 1 0.34 0.590 1.2 0.38 0.602 1.2 0.64 0.486 Heteroponera imbellis (CCS) 12 1 0.74 0.438 1.5 1.52 0.283 1.2 2.69 0.164 Hypoponera sp.1 (C) 30 1 0.30 0.614 2.0 0.01 0.989 1.2 0.01 0.948 Lordomyrma QM1 (TCS) 6 1 1.05 0.363 1.4 1.01 0.391 2.0 0.58 0.581 Mayriella abstinens complex (TCS) 8 1 3.31 0.143 1.6 1.53 0.279 1.8 2.95 0.119 Monomorium tambourinense (CCS) 9 1 0.06 0.813 1.2 0.53 0.532 1.9 0.46 0.633 Pheidole QM1 (GM) 4 1 0.00 1.000 2.0 0.21 0.815 2.0 0.71 0.522 Pheidole QM2 (GM) 11 1 1.28 0.321 2.0 0.26 0.774 2.0 0.31 0.741 Pheidole sp.2 (GM) 5 1 0.04 0.854 1.4 0.74 0.475 1.6 1.63 0.264 Rhytidoponera victoriae (O) 12 1 5.47 0.079 1.7 0.71 0.505 1.7 2.07 0.200 Solenopsis QM1 (C) 8 1 0.21 0.672 2.0 0.05 0.951 2.0 1.89 0.213 Strumigenys deuteras (C) 4 1 4.57 0.099 2.0 0.29 0.759 2.0 0.29 0.759 † Number of plots (N = 30) where that taxon was present. Degrees of freedom are corrected by Huynh-Feldt Epsilon values. Ant functional groups are shown in parenthesis; C, Cryptic species; CCS, Cold climate specialists; GM, Generalised Myrmicinae; O, Opportunists; TCS, Tropical climate Specialists.

Coarse arthropods

a Total abundance b Taxon richness c Araneae 200 16 10 180 15 8 160 14 6 140 13 4 120 12 100 11 2

Abundance/taxon richness richness Abundance/taxon 80 10 0 0.5 Before 1.5 After 2.5 After 3.5 0.5 Before 1.5 After 2.5 After 3.5 0.51Before . 52 After .53 After .5 3 days 3 months 3 days 3 months 3 days 3 months

d Coleoptera e Diplopoda f Diptera 50 25 3

40 20 2 30 15 20 10

Abundance Abundance 1 10 5

0 0 0 0.51.52.53.5 Before After After 0.5 Before 1.5 After 2.5 After 3.5 0.5 Before 1.5 After 2.5 After 3.5 3 days 3 months 3 days 3 months 3 days 3 months

g Heteroptera h Isopoda i Pauropoda 16 50 12 40 10 12 8 30 8 6 20 Abundance 4 4 10 2 0 0 0 0.5 Before 1.5 After 2.5 After 3.5 0.5 Before 1.5 After 2.5 After 3.5 0.5 Before 1.5 After 2.5 After 3. 3 days 3 months 3 days 3 months 3 days 3 months

Ant species j Total abundance k Species richness l Hypoponera sp.1 60 9 50 50 8 40 40 7 30 30 6 20 20 5 10 4 10

Abundance/species richness Abundance/species 0 3 0 0.5 Before 1.5 After 2.5 After 3.5 0.51Before . 52 After .53 After .50.5 Before 1.5 After 2.5 After 3.5 3 days 3 months 3 days 3 months 3 days 3 months

Figure 7.2 Total abundances, taxon richness, and abundances (mean, SE) of individual arthropod taxa/species before and after the herbicide application for the control (open circle) and treated (closed square) plots. Figure shows only common taxa/species that occurred in all the plots before application, or that showed significant responses to the experimental treatments (P < 0.05) (see also Table 7.3).

137

7.4 Discussion

Effects of the glyphosate herbicide under field conditions Under field conditions, application of herbicides often mediates assemblage compositions of groups of animals, such as birds, small mammals and invertebrates (Sullivan 1985; House et al. 1987; Brust 1990; Haughton et al. 1999b; Haughton et al. 1999a; Taylor et al. 2006). Responses to the herbicide applications are, however, generally attributable to indirect effects caused by the death of vegetation. If one wishes specifically to assess the effect of herbicide (unconfounded by other mechanisms), indirect effects caused by the loss of vegetation must be separated from impacts. The present study assessed effects of glyphosate herbicide under dense canopy cover with sparse understorey vegetation so that the direct effects of the herbicide were detectable. Furthermore, the dense canopy cover and surrounding vegetation of the study sites reduced fluctuations in microclimatic conditions (e.g. temperature, moisture content) that otherwise may have caused changes in arthropod assemblages regardless of the herbicide application (Lindsay & French 2004). In the absence of indirect impacts, the results suggested that application of the glyphosate formulation had negligible impacts on the assemblages of soil and litter macro-arthropods in the floor of subtropical rainforest, in spite of high application rate.

As has been seen in ecotoxicological studies of some other pesticides (i.e. insecticides, fungicides), loss of or reduced toxicity of the pesticides is not uncommon when tested under field conditions, particularly if that pesticide exhibited only slight to moderate toxicity under laboratory conditions (Hassan et al. 1988). Under laboratory conditions, glyphosate is weakly toxic to the species of the epigaeic arthropods that have been tested (viz. Isopoda, Araneae) (no more than 50% reduction in their survival or longevity, Eijsackers 1985; Mohamed et al. 1992; Haughton et al. 2001b). In addition to their inherently low toxicity, glyphosate herbicides are likely to be inactivated in the field by adsorption to soil and by decomposition through the activities of micro-organisms. In my study, adsorption of glyphosate may have been particularly important, as the experimental plots were located on basaltic soil with a high content of clay and organic matter (Hubble et al. 1983) which are known to adsorb considerably more glyphosate than other sandy minerals (Sprankle et al. 1975a; Glass 1987).

138

One may argue, however, that the size of my experimental plots was so small (3 x 3 m quadrat) that recolonising arthropods from the surrounding habitat may have supplemented the loss of arthropods caused by the herbicide. The present study employed litter extraction that primarily sampled the less mobile arthropods (cf. pitfall traps which primarily sample active epigaeic arthropods, Majer 1997; Bestelmeyer et al. 2000), thereby minimising the confounding effects of arthropod recolonisation over the duration of the field experiment. Furthermore, if loss and subsequent recovery of arthropods had occurred as a result of herbicide application, arthropods with different vagilities and reproductive abilities should have recolonised at different rates (den Boer 1990) causing changes in arthropod assemblage compositions over time between control and treated plots. This was not observed in my results.

Folgarait (1998) argued that ants may not respond to the presence of pollution if they nest away from the influence of pollutants (e.g. species having subterranean or arboreal colonies) or if they actively change foraging behaviours in order to avoid contact with pollutants (e.g. epigaeic foragers). Unlike pitfall trapping that generally collects ants with the above mentioned characteristics, litter extraction primarily collects small, slow moving cryptic species (Andersen 1995a). These include Carebara QM1, Discothyrea QM1 and Hypoponera sp.1, which nest and forage in forest litter or in surface soil where they may be reached by herbicide residues (Shattuck 1999). Over 60% of my ant catches came from cryptic species (563 individuals) and lack of response from these ant species may indeed indicate that the toxicity of the herbicide was negligible.

The relatively small sample sizes generated in this study do increase the potential for Type II errors. Like the results of all statistical tests, my results merely indicate that the null hypothesis of non-significance is tenable (Snedecor & Cochran 1980). However, given the large number of non-independent statistical tests in Tables 7.1, 7.2 and 7.3, the lack of significant results gives me added confidence in my conclusions.

Management implications When used periodically (e.g. six-monthly or annually), glyphosate in the form of Roundup® Biactive™ appears to be suitable for the control of unwanted plants in rainforest restoration sites without detrimental impacts on macro-arthropods. Some types of glyphosate formulations may exhibit greater toxicity to non-target organisms. Glyphosate comes in a variety of commercially available formulations that contain

139

different surfactants and other by-products (Giesy et al. 2000). For example, polyethoxylated tallow amine (POEA) is the surfactant used in some commercially available glyphosate formulations (e.g. Roundup®). Although little is known about its toxicity to soil and litter macro-arthropods, POEA is known to be considerably toxic to a number of other organisms (e.g. rats, amphibians, aquatic invertebrates, see Giesy et al. 2000; Relyea 2005). In this study I used Roundup® Biactive™ whose formation contains a surfactant (undisclosed component) that is substantially less toxic to the organisms that were found susceptible to POEA (Mann & Bidwell 1999; Tsui & Chu 2004). Although other types of glyphosate herbicide may be cheaper and more effective than Roundup® Biactive™, these alternatives must be used with caution, as their potential field toxicities have not been evaluated.

Furthermore, the toxicity of glyphosate herbicides may vary according to different soil conditions and/or restoration practices. Certain types of soil (e.g. sandy loam) and the presence of phosphate fertilizer are known to reduce or inhibit the adsorption of the herbicide to soil particles, consequently increasing the persistence of residual glyphosate in soil (Sprankle et al. 1975a; Hance 1976; Eberbach & Douglas 1983; Glass 1987). Further studies are warranted to establish the effects of different substrate conditions on the toxicity of glyphosate herbicides and the impact of other herbicide formulations commonly used in forest restoration.

140

8 GENERAL DISCUSSION

8.1 Summary and key findings of the data chapters (Chapters 3-7)

The present study first carried out a baseline survey to investigate the impacts of rainforest clearance, and associated subsequent land use as pasture, on assemblages of soil and litter arthropods (Chapter 3). Responses of arthropods were analysed using various taxonomic and functional groups, namely ‘coarse’ arthropods (all macro-arthropods sorted to Order/Class), ant genera, ant species, ant functional groups and ant biogeographical affinities. Regardless of arthropod group (with the exception of ant geographical affinities) or sampling method used (pitfall trapping or leaf litter extraction) there were clear differences in the composition of arthropod assemblages between rainforest and pasture. The sensitivity of the assemblage response, however, did not show a progressive increase with finer levels of taxonomic resolution: ant species best segregated the two habitat types, followed by coarse arthropods and ant genus datasets.

Individual arthropod taxa were assessed for their use as bio-indicators of habitat change using an Indicator Values protocol (Dufrene & Legendre 1997, see also Chapter 3). There were more arthropod taxa characteristic of rainforest than of pasture. More habitat ‘increasers’ (those which were found in both habitat types but were significantly more abundant in one) were identified when arthropods were sorted to coarser taxonomic resolutions, whereas more habitat ‘specialists’ (those which were found exclusively in their associated habitat) were found from the species-level dataset. However, the use of habitat indicators, especially ant species, may be unreliable due to their patchy distributions. To overcome this problem, ‘composite rainforest/pasture indices’ were generated by combining information from sets of indicator taxa. The data obtained from the baseline survey were incorporated into the analyses of the manipulative field experiments (Chapters 4, 5 and 6).

For all subsequent chapters the present study has adopted an experimental approach. Chapter 4 investigated the effects of distance from rainforest remnants on the recolonisation patterns of soil and litter arthropods in ‘restored’ habitat patches within a pasture matrix. The efficacy of inoculation (translocation of small quantities of litter which contained live arthropods from rainforest to restored habitat patches) was also tested for its potential to boost the rate of arthropod colonisation. After nine months, there

141

was little colonisation by rainforest-dependent taxa in any of the experimental plots beyond those closely adjacent to remnant forest patches (Table 8.1). Inoculation was unsuccessful in increasing the extent of arthropod establishment: the majority of rainforest arthropods from the raw inoculum failed to persist within the experimental plots. A number of explanations that may potentially account for the observed results are suggested. Although an experimental approach provides an opportunity to test explicitly for factors considered important for the development of biota in restored habitat patches, limitations associated with the experimental design affected the interpretation of the results.

Chapter 5 assessed the effects of mulch quality and quantity on the colonisation patterns of soil and litter arthropods. Experimental habitat patches were established by adding sterilised hay (a conventionally used mulching material) or forest woodchip mulch (a structurally more complex alternative) at two depths (shallow 3-5 cm and deep 10-15 cm) to simulate conditions during the initial stages of rainforest restoration. The habitat patches were all unshaded, and were positioned adjacent to the edges of rainforest remnants to minimise the effect of isolation (Table 2.1). It was expected that woodchip mulch would be colonised by more rainforest arthropods than hay mulch. Despite its simpler composition, hay performed better than woodchip mulch in facilitating colonisation by arthropods characteristic of rainforest (Table 8.1). Shallow hay was particularly associated with rainforest-dependent ant species, although these species may have utilised the experimental plots only for foraging, with their colonies established elsewhere. Unlike the results of the present study, other ecological studies have suggested that species and structural complexities of litter may be important for the establishment of diverse soil and litter arthropod assemblages (Table 1.5, see also Wardle 2006). These studies, however, did not pay particular attention to the initial stages of restoration where no other physical structures (i.e. established trees) are available to shelter arthropods. The present study (Chapter 5) was carried out in this context, and found that characteristics of litter (e.g. levels of insulation, nutrient content) may be more important than litter complexity per se, supporting the use of (primarily) monocultural hay during planting activities of rainforest restoration. However, neither hay nor woodchip mulch impeded colonisation by arthropods characteristic of pasture.

142

Table 8.1 General patterns in the levels of composite rainforest/pasture indices that were found in factorial analyses of the manipulative field experiment (Chapters 4, 5 and 6). Levels of composite indices under different experimental treatments are represented comparatively by “>”, “<” or “=” symbols. The results are presented only for composite habitat indices that showed statistically significant responses (P < 0.05).

Chapter Chapter 4 Chapter 5 Chapter 6 Factors tested Distance Mulch quality Mulch depth Shade Mulch depth Coarse arthropods Rainforest indices (pitfall traps) -30m > 0m > 15m = 100m = 400m ns ns 0% < 90% < 50% ns Rainforest indices (litter extraction) ns Hay > Woodchip ns ns ns

Pasture indices (pitfall traps) ns ns ns 0% > 50% > 90% ns Pasture indices (litter extraction) - - - - -

Ant species Rainforest indices (pitfall traps) -30m > 0m > 15m = 100m = 400m Shallow hay > other treatments† ns Shallow > Deep Rainforest indices (litter extraction) ns - - - - Pasture indices (pitfall traps) ns ns ns 0% > 50% > 90% ns Pasture indices (litter extraction) ns ns ns - - ns: Not statistically significant. - : Composite habitat indicator was not calculated due to small number of component taxa (less than two) sampled. †: Interaction effect between mulch quality and depth was significant.

143

Chapter 6 assessed the effects of varying degrees of shading (0%, 50% or 90% shadecloth) and depth of woodchip mulch (shallow 3-5 cm or deep 10-15 cm) on the colonisation patterns of soil and litter arthropods. The influence of shading has important implications for the management of rainforest restoration, as different restoration techniques may vary greatly in their canopy cover during the early stages of restoration (ca. 5-22 years since reforested, Kanowski et al. 2003). An increased amount of forest litter was expected to offset the deleterious effects of sparse canopy cover. As in Chapter 5, all experimental plots were established adjacent to the edges of remnant rainforests. Both 50% and 90% shading encouraged colonisation by coarse arthropods characteristic of rainforest (Table 8.1). Shallow mulch positively affected colonisation patterns of rainforest ant species. The extent of colonisation by pasture-associated arthropods declined progressively with increased shading. The results suggest that more widely-spaced plantations (such as timber plantations) may facilitate some colonisation by rainforest arthropods. However, in order to suppress the extent of invasion by pasture-associated arthropods, it may be necessary to establish a fully closed canopy (such as that achieved rapidly in densely-spaced restoration plantings). Changes in the levels of canopy cover (shading) consistently influenced organisation of arthropod assemblages in both the present study and other restoration studies summarised in the literature review (Table 1.5). This confirms that canopy cover is indeed one of the primary attributes influencing colonisation patterns of arthropods in restored vegetation, and this should be taken into consideration when designing restoration programs. It should be noted, however, that the establishment of canopy cover does not necessarily exert a positive influence, as other types of habitat may require moderate or sparse canopy cover, and fully established canopy cover may even cause deleterious impacts on successful colonisation by native fauna (e.g. restoration of open eucalyptus woodland, see Andersen 1993).

Both Chapter 5 and 6 suggested that the effect of litter depth (3-5 versus 10-15 cm) may be of subsidiary importance to other factors such as shading and litter quality. The results are broadly parallel with those of other studies of active-restoration that often failed to find significant responses of arthropod colonisation to different amounts of litter (Table 1.5). Furthermore, as far as ants are concerned, the present study suggested that shallow mulch, that simulated the depth of naturally occurring rainforest litter (ca. 3 cm), may be

144

more suitable for rainforest species, provided that other factors (e.g. temperature regime) are favourable.

Chapter 7 describes a separate field experiment that investigated the effects of glyphosate herbicide (Roundup® Biactive™) on soil and litter arthropods inhabiting the litter of existing remnant rainforest sites. Arthropods were extracted only from litter, as this sampling method primarily collects less mobile arthropods (cf. pitfall traps) and was, therefore, expected to best detect the effects of the herbicide within the limited area (3 x 3 m quadrat) of the experimental plots. Despite the high application rate, there were neither immediate nor delayed impacts of the herbicide on the composition of soil and litter macro-arthropod assemblages. The inherently low toxicity of the herbicide formulation and its adsorption to soil particles may have contributed to the results. The present study suggested that the use of glyphosate herbicide formulated as Roundup® Biactive™ is suitable for the control of unwanted plants in rainforest restoration sites without deleterious impacts on soil and litter arthropods, from the perspective of arthropod biodiversity.

8.2 Experimental approaches in restoration studies

The restoration of disturbed habitats provides an opportunity to conduct ‘acid tests’ (Bradshaw 1987) of our understanding of ecosystem recovery, which also provide information needed to improve restoration techniques and management. However, restoration ecology has chiefly relied upon post-hoc empirical observation of actual restoration programs (see Section 1.3.3), which has hindered systematic examination of individual factors. The advantages of an experimental approach are exemplified by a comparison between one component of the manipulative field experiment (Chapter 6) and a study of actual rainforest restoration in the same region (Nakamura et al. 2003). The latter study suggested that an extensive canopy cover was important for the development of a diverse assemblage of rainforest arthropods; however, interpretation of the results was limited due to the presence of confounding factors and the lack of replication. In contrast, in the field experiment (Chapter 6) replicated experimental units were constructed in which the degree of shading (an important attribute associated with canopy cover) was manipulated, while other extraneous factors were controlled. The field experiment reinforced the importance of shading for successful colonisation by

145

forest-dependent arthropods and also provided further information on the effects of varying degrees of shading.

Furthermore, this approach provides a cost-effective way of trialing aspects of restoration techniques. If practitioners try out very novel restoration techniques, the results will inform future restoration programs (i.e. 'trial-and-error'/adaptive management approach, Cabin 2007). However, at real-world spatial scales this may not only exhaust limited budget and supply of resources but may cause further degradation of habitats being restored. With relatively small costs and resources, the manipulative field experiment simulated conditions potentially experienced by colonising organisms, providing information about possible consequences that may result from various restoration techniques and management that have yet to be implemented.

Despite the advantages of an experimental approach, there are unavoidable constraints associated with the design and scale of experiments. These constraints were particularly apparent in the tests of habitat isolation and inoculation (Chapter 4) where the majority of rainforest-dependent arthropods failed to establish in experimental plots located within pasture (i.e. the 15, 100 and 400 m treatments). As there was no significant effect of distance, it was impossible to determine whether the lack of colonisation in the isolated plots was due to the genuine effects of isolation or other factors closely linked to the experimental design (i.e. habitat quality, small size of the plots, lack of recognisable habitat, short duration of the experiment, see the Discussion in Chapter 4). Consequently the inconclusive results were deemed to have limited implications for actual restoration efforts. In other components of the manipulative field experiment (Chapters 5, 6), experimental habitat patches were colonised by rainforest arthropods, and the experiment revealed the effects of the selected factors (mulch type, mulch depth, shading) on colonisation patterns of arthropods. Interpretation of the results was, however, still limited due to aspects of the experimental design, such as the quality of the created habitat patches and spatial and temporal scales of the experiment. Unlike the manipulative field experiment, the herbicide field experiment (Chapter 7) utilised natural habitat, eliminating problems associated with the artificiality of created habitat patches. Once again, issues associated with the scale of the experiment (both temporal and spatial) were of concern, and the observed results were interpreted carefully in this context.

146

A more productive experimental design may be to implement controlled and replicated restoration efforts over a larger area, in collaborations between researchers and practitioners. This would alleviate most of the problems associated with the manipulative field experiments, as actual restoration programs are implemented under controlled conditions with realistic time- and spatial-scales. Nevertheless, this kind of approach would be expensive and would take a prolonged time for the experimental sites to achieve an age at which habitat components (such as shading) had developed sufficiently to allow meaningful measurements of the arthropods’ responses. Further, collaboration between these two groups may be challenging because the goals and practices of researchers often conflict with those of restoration practitioners (Cabin 2007).

This project used a feasible, cost- and time-effective experimental approach which generated new information to inform future efforts in rainforest restoration. However, improvement in the design of the manipulative field experiment is needed if similar approaches are to be adopted for future studies. In order successfully to test the effect of habitat isolation (Chapter 4), for example, it would have been useful first to demonstrate that conditions of the created habitat patches were suitable for potentially-colonising target fauna. This could be done by iteratively improving habitat conditions within the plots until translocated organisms established and persisted in plots which were located within a matrix habitat (cf. Davis & Jones 1986). However, this may require considerable time and effort. Other unavoidable limitations of the field experiment (e.g. its spatial and temporal scales) may be best dealt with by the use of parallel post-hoc empirical studies of actual restoration, whose findings may be compared with those of field experiments in order to test the generality of the conclusions.

8.3 Use of reference information in restoration studies

The results of the baseline survey (Chapter 3) provided reference information that described the distribution of soil and litter arthropods in undisturbed (rainforest) and disturbed (pasture) habitats. Reference information was essential for the interpretation of the data obtained from the manipulative field experiment because the distribution of soil and litter arthropods was poorly known in the study area, as in most regions of the world (but with a limited number of exceptions, see Dunger et al. 2001). The absence of reference information may be justified in some cases of restoration studies, with the assumption that disturbed sites are presumably devoid of most organisms (as in minesite 147

restoration), so that conventionally-used diversity indices, such as species richness and Shannon’s index, can be used validly to measure the recovery of restored sites from young (no organisms) to old, developed stages (diverse array of organisms). An increase in the values of these indices, however, does not necessarily point to the development of organisms characteristic of undisturbed habitats. Further, the use of these diversity indices may be entirely pointless in the studies of old-field restoration, as both disturbed (e.g. pasture) and undisturbed habitats often contain diverse arrays of arthropods, but their composition may be clearly different (as in Chapter 3, see also Appendix 4a).

The baseline survey was unique in the level of spatial replication of reference sites (N =12), as other restoration studies have generally used few spatially replicated reference sites (see Table 1.4). This allowed the establishment of reliable reference information, as arthropods were sampled from a relatively large number of sites across the study region to encompass the natural heterogeneity of arthropod distributions, which is expected to occur even within the same type of reference habitat. This not only provided comprehensive information on arthropod assemblages typically found in the reference habitats of this study region, but also yielded a large number of habitat indicator taxa that were used in the subsequent studies. If a smaller number of spatial replicates had been used in this study, the number of habitat indicators would have been greatly reduced because many of the habitat indicator taxa, particularly of habitat ‘specialists’, occurred at less than half of the 12 sites sampled.

Extensive spatial replication was, however, implemented at the expense of temporal replication: arthropods were collected in the baseline survey of reference sites using one-off sampling that took a ‘snap-shot’ of assemblage composition in one particular season. Limited data on both short- (i.e. seasonal variation) and long-term (i.e. inter-annual variation) variability poses questions about the reliability of the reference information for use in the following experiments, which sampled arthropods at different times from the baseline survey. Nevertheless, the effect of temporal variability may not be a significant problem in the present study because a) the effect of seasonal (short-term) variation was controlled by conducting all sampling in the same season (summer), and b) long-term variability may not be of significant concern because all arthropod sampling was carried out within a two-year period, under the influence of similar climatic regimes (both years slightly drier than long-term average rainfall, see Section 2.1). It should be noted, however, that the generality of reference information obtained in the present study 148

is unknown, given that most ecosystems are dynamic, changing their characteristics as a result of a wide range of temporal environmental changes (Hobbs & Harris 2001; Harris et al. 2006).

Unless information related to taxonomy, ecology and distribution of individual soil and litter arthropods is well established through repeated comparison of reference sites in different years, it may be best to collect contemporary reference information each time a restoration study is carried out. Furthermore, long-term temporal replicates do become a critical component of the reference information if one wishes to carry out a longitudinal survey of rainforest restoration spanning over several or more years.

8.4 Choice of target taxa, measurements and methods for monitoring the state of soil and litter arthropods in restored rainforests

Given the complexity of rainforest ecosystems, there is a myriad of biotic and abiotic variables which potentially can be measured to monitor the state of rainforest restoration. In addition to their intrinsic values, soil and litter arthropods possess characteristics that make them suitable for use as bioindicators of habitat recovery (Section 1.3.1). Even within this focal group, however, there are various options available for monitoring, and we must choose a set of response variables as well as sampling methods that are likely to best reflect the state of arthropod assemblages. Here I summarise the patterns of arthropod assemblages measured by different sampling, sorting and analytical options used in this project (viz. sampling methods, taxonomic resolutions, ant functional groups, composite habitat indices) and explore their potential use for effective monitoring in the context of rainforest restoration.

Sampling methods The baseline survey showed that both pitfall trapping and litter extraction yielded many of the rainforest/pasture indicator taxa, and clearly separated rainforest from pasture in terms of the composition of arthropod assemblages. The same, however, was not true for the manipulative field experiment: most of the significant responses were found from pitfall-trapped arthropods (e.g. Table 8.1). As has been supported by other restoration studies (see Section 1.3.3), this study’s results also suggested that pitfall traps may be used effectively to provide information on the development of colonising arthropods in restored rainforests under different habitat conditions.

149

It should be noted, however, that the results collected by pitfall trapping alone may not necessarily represent the whole suite of soil and litter arthropods, as this method only collects active epigaeic arthropods. Litter extraction, on the other hand, is known to collect different groups of less mobile, cryptic arthropods that live in soil and litter. Supporting this, almost all of the rainforest indicators of ant species unique to litter extraction were, in fact, soil and litter dwelling species (viz. Carebara QM2, Discothyrea QM1, Hypoponera QM5, Hypoponera sp.1, Lordomyrma QM1, Mayriella abstinens complex, Strumigenys harpyia, see Appendix 3b). This project demonstrated that a combination of complementary sampling procedures (e.g. pitfall traps and litter extraction with Tullgren/Berlese funnels or Winkler sacks, see also Delabie et al. 2000) can be used effectively when investigating sites with various habitat complexities (e.g. sites with bare ground through shallow to deep woodchip or hay mulch) typically found in forest restoration where one sampling method alone cannot adequately collect all components of soil and litter arthropods.

Taxonomic resolution The baseline survey compared sensitivities of different taxonomic resolution (viz. coarse arthropods, ant genus, ant species) to contrasting changes in habitat conditions (rainforest versus pasture). It is not surprising that the species-level analysis produced the most sensitive results as this approach reflected habitat preferences of individual species. Species-level sorting, however, does require extensive taxonomic knowledge which is often unavailable for many groups of arthropods. In order to evaluate the state of the whole soil and litter dwelling macro-arthropod assemblages, coarser taxonomic resolutions were also employed in this project. However, this approach may mask responses of certain sensitive species by other, less-sensitive species. Thus, even if the higher-level composition of arthropod assemblages in the restored habitat resembles that of the undisturbed habitat, the underlying species-level composition of the two habitats may not necessarily be similar. Supporting this, the restoration studies that employed coarse taxonomic resolutions showed assemblage compositions similar to those of undisturbed reference sites, indicating the ‘success’ of the project within one or two decades of restoration (e.g. Jansen 1997; Nakamura et al. 2003); however, this was not true for most of the studies that employed species-level sorting even when restored sites were over a few decades old (e.g. Bisevac & Majer 1999; Reay & Norton 1999; St. John

150

et al. 2002; Redi et al. 2005; Ottonetti et al. 2006; Grimbacher et al. 2007; but with some exceptions, see Simmonds et al. 1994; Andersen et al. 2003).

Ideally, the entire suite of soil and litter arthropods should be sorted to species, but this is not feasible unless there are unlimited amounts of time and resources available for identification and taxonomic studies (Section 1.3.2). This study’s combination of higher-taxon sorting of all arthropods, together with species-level sorting of a significant major taxon (ants), is a feasible compromise between comprehensiveness and detail in monitoring responses of arthropods in restored rainforests. This project demonstrated that observed similarities in overall patterns in both coarse arthropods and ant species reinforced conclusions on the general patterns of arthropod colonisation in response to changes in habitat conditions (see Table 8.1). In addition, some of the unique responses observed in the ant species dataset provided additional insight into colonisation mechanisms of this particular arthropod group (Table 8.1, see also Chapters 5 and 6).

Ant functional groups Hoffmann and Andersen (2003) reviewed Australian studies of ant functional groups and suggested that their organisations were affected generally by changes in habitat characteristics, such as the amount of insolation and leaf litter. The results of this study’s baseline survey broadly concurred with their findings: relative abundances of Generalised Myrmicinae and Specialist predators increased in the rainforest, presumably because of the presence of canopy trees and ample litter. Opportunists, known as a strong disturbance ‘increaser’ group that often show opposite responses to those found in Generalised Myrmicinae (Hoffmann & Andersen 2003), dominated in pasture (Chapter 3, see also Appendix 4b). While the baseline survey clearly demonstrated their responses to changes in habitat conditions, this was not true for the manipulative field experiment.

The lack of significant responses at functional group level in the field experiment may, in part, be attributable to the generality of the scheme which is not designed for detailed studies of local assemblage structures within small-scale habitat patches (Andersen 1997a). In addition, inadequate sampling intensity may have failed to detect responses of the less abundant functional groups (viz. Specialised predators). Further, the observed results may be explained by the lack of understanding of the ecology and behaviour of ant communities in this region. The ant functional group scheme was primarily developed on the basis of the ecological and behavioural responses of ants in areas other than tropical

151

and subtropical climatic regions (Greenslade 1978; Andersen 1995a; Andersen 1997b). Hence species-level ecological and behavioural studies need to be carried out across the various habitats of the particular bioregion being studied before ants can be reliably classified into their respective functional groups (Bestelmeyer & Wiens 1996; Ottonetti et al. 2006). With the limited and regionally biased knowledge currently available, this approach, at present, may provide only coarse predictive power in this study region.

Composite habitat indices Patchiness of species distribution is an unavoidable constraint associated with the study of arthropod assemblages (Giller 1996; Veech et al. 2003). Even within the same type of habitat with apparently homogeneous environmental conditions, soil and litter arthropod taxa occurred in a patchy fashion, which impeded the generation of reliable habitat indicator taxa with high habitat specificity and fidelity (Chapter 3). The manipulative field experiments demonstrated that the utility of individual indicator taxa was, in fact, limited as most indicator taxa exhibited non-significant responses to the experimental treatments. A lack of reliable habitat indicators inhibits our understanding of the colonisation patterns of desired/undesired arthropod groups (i.e. taxa characteristic of either rainforest or pasture) under different environmental conditions. Use of individual indicator taxa might have been more reliable if sampling replication and intensity were increased within both the baseline survey and the field experiment; however, in reality, this may not be a feasible option in most cases of restoration studies due to limited resources. In the present study, the use of composite habitat indices successfully quantified the extent to which a site is rainforest-like or pasture-like, in terms of its arthropod assemblage composition, overcoming the limitations associated with the separate use of individual indicator taxa.

However, this approach would be of limited use if there was only a small number of component taxa (< 2 taxa) comprising a composite habitat index. This constraint was apparent in the case of litter-extracted arthropods from which only one component taxon was occasionally sampled (some of the indices were, therefore, not calculated, see Table 8.1). For this approach to be used effectively, reference information needs to yield at least two component indicator taxa (and preferably more).

152

8.5 Introduced species in restored rainforests

Invasion of exotic species is a worldwide phenomenon (Olden 2006), and interactions between native and exotic species have been considered as one of the key processes influencing the successful development of ecological restoration (Suding et al. 2004). In particular, competitive exclusion of native arthropods by exotic ants is well documented in Australia and elsewhere (McGlynn 1999, see also references therein), and their potential impacts on the development of arthropods in restored habitat patches needs special attention. In the study of restored dune forests in Australia, for example, native ant species appeared to be displaced by the super-abundant exotic ant species, Pheidole megacephala, in old, well-established restoration sites (Majer 1985). Pheidole megacephala is a highly invasive species, that is capable of dramatically altering assemblages of native ants and other arthropods (Heterick 1997; Hoffmann 1998; Vanderwoude et al. 2000b; Ness & Bronstein 2004). In the baseline survey, P. megacephala was found in one of the apparently undisturbed rainforest remnants, contributing over 70% of the total ant catch in this site (site S12, Appendix 2, see also Chapter 3). This species was also found in one of the two experimental plots (site D1) that were established in relatively close proximity (400 m) to the infested rainforest (other experimental plots were established further away from the infested rainforest and were adjacent to other rainforest remnants with no observed infestation). Due perhaps to their susceptibility to desiccation (Hoffmann 1998), P. megacephala was not found in pasture, suggesting that P. megacephala may be able to disperse and invade restored rainforest patches if they are established near infested habitats. A previous survey of actual rainforest restoration found no infestation by P. megacephala in this region (Nakamura et al. 2003); however, another survey conducted in tropical north Queensland found P. megacephala in a newly restored rainforest site (King et al. 1998). Although potential pathways and impacts of many exotic species are largely unknown, it is important for restoration practitioner to avoid accidental transfer of exotic species in the restored habitat patches (e.g. through mulch transport, seedling plantings with untreated soil).

8.6 Recommendations for restoration practice

Clearing of rainforest and subsequent land use for pasture replaces forest-dependent arthropods with different sets of arthropods, including many species typical of highly disturbed habitats (e.g. Cardiocondyla nuda and Rhytidoponera metallica in this study,

153

see Chapter 3). This warrants the restoration of cleared rainforest in order to recover diverse arthropod assemblages that existed before anthropogenic disturbances. From the point of view of soil and litter arthropod biodiversity conservation, Tables 8.2 and 8.3 summarise recommendations that should be incorporated into, or at least taken into consideration during restoration programs in rainforest landscapes. They are of two sorts: recommendations for effective monitoring of soil and litter arthropods in restored rainforests (Table 8.2), and improvements in restoration management designed to facilitate colonisation by arthropods characteristic of rainforests (Table 8.3).

Table 8.2 Information to aid decisions about monitoring the development of soil and litter arthropod assemblages in rainforest restoration. Issues Information to aid decisions Reference information Establishment of reference information is a critical part of monitoring programs, as this provides milestones against which the development of fauna can be evaluated. Types of reference Two types of reference information are useful, representing the information states that: a) restoration projects aim to ultimately achieve (i.e. undisturbed system), and b) existed before restoration commenced (i.e. disturbed system). Spatial replication of Spatial replication is needed to encompass the heterogeneous reference information distributions of organisms within a habitat type.

Temporal replication of Temporal replication may be important particularly for reference information longitudinal monitoring spanning several years or more.

Use of pitfall trapping Pitfall trapping can be used to effectively monitor the state of active ground-foraging arthropods in restored rainforests provided that the sampling protocol is standardized across sites. Additional trapping An additional method that samples other components of methods arthropods, such as litter extraction designed to sample less mobile litter fauna, is useful to complement pitfall data. Taxonomic resolution Species-level sorting of the entire arthropod assemblages is an impractical option in most monitoring programs. A combination of a higher-taxon approach that sorts all of the macro-arthropods at coarse taxonomic resolution (i.e. Order), and a species-level approach that focuses on particular group of arthropods (e.g. ants) may provide a feasible option to best reflect the state of soil and litter arthropod community in restored rainforests. Composite habitat Due to patchy species distributions of organisms, the usefulness of indices individual habitat indicator taxa tends to be limited. The use of composite habitat indices may alleviate this problem.

154

The central aim of this thesis was to explore different ways to maximise the biodiversity values of restored land through successful colonisation by arthropods characteristic of rainforest. Augmentation of the biodiversity value, however, represents only one of the goals available for restoration programs (e.g. restoration of ecological function, timber production, carbon sequestration, aesthetic improvement, rescuing particular endangered species). As these goals are often not compatible with each other (Catterall et al. 2005; Lamb et al. 2005), prioritisation of these goals is required if more than one goal is incorporated into a restoration program. For example, restoration practitioners aiming to gain economic benefits in timber plantations may need to trade-off biodiversity values to achieve both outcomes (i.e. higher density planting to increase the biodiversity values may negatively affect timber production). With increasing interest in certification of timber production practices for commercial advantage (Gullison 2003; Karna et al. 2003), it is hoped that augmentation of biodiversity values will become a more feasible and favourable option for not only ecologically-oriented restoration but also other types of restoration programs (e.g. economically-oriented plantation) in the near future.

155

156

Table 8.3 Information to aid decisions in rainforest restoration management, based on the findings of this project. Relevant Factors chapters Information to aid decisions Habitat isolation Chapter 4 Restoration sites close to existing rainforest remnants are likely to more rapidly acquire rainforest soil and litter arthropods, because they are likely to have limited dispersal abilities. Further studies of their ability to colonise isolated restored habitat patches are needed. Inoculation of arthropods Chapter 4 Translocation of small quantities of rainforest litter (containing live arthropods) to isolated habitat patches may be of limited use. Modification of the inoculation procedure (factors such as, timing, handling, quantity, and selective collection of inoculum) may improve the efficacy of the process. Mulch quality (hay vs Chapter 5 During the initial stages of reforestation (i.e. no shading), hay mulch perform better than woodchip woodchip mulch) and depth mulch in facilitating colonisation by arthropods characteristic of rainforest; however, neither hay nor woodchip mulch inhibits arthropods invading from surrounding pasture without the provision of adequate shading (see below). Shallow (3-5 cm) hay was preferred by ants characteristic of rainforest; however, other groups of arthropods (e.g. Coleoptera, Isopoda) were associated with deep hay (10-15 cm). The optimum amount of hay may vary among different groups of arthropods. Shading and mulch depth Chapter 6 Moderate levels of canopy closure (i.e. 50% shading), such as that produced by tree spacing typical of timber plantations, may be sufficient to facilitate colonisation of reforested land by rainforest arthropods. However, greater canopy shading (90%) would be needed to inhibit re-invasion by arthropods from surrounding pasture habitat. There was no evidence that the addition of extra mulch compensated for reduced levels of shading. Herbicide application Chapter 7 The use of glyphosate herbicide in the form of Roundup® Biactive™ is suitable for control of unwanted plants without deleterious impacts on soil and litter arthropods. Introduced species Chapters 3, 4 Continuous monitoring is needed to ensure that the restored rainforest is free from invasion by exotic arthropod species (e.g. Pheidole megacephala), and care is needed to avoid accidental transfer of exotic species into restored habitats.

REFERENCES

Adam, P. (1992) Australian Rainforest. Oxford University Press, Oxford.

Addo-Bediako, A., Chown, S.L., & Gaston, K.J. (2000) Thermal tolerance, climatic variability and latitude. Proceedings of the Royal Society of London. Series B, 267, 739-745.

Adis, J. (1979) Problems of interpreting arthropod sampling with pitfall traps. Zoologischer Anzeiger, 202, 177-184.

Agosti, D., Majer, J.D., Alonso, L.E., & Schultz, T.R., eds. (2000) Ants: Standard Methods for Measuring and Monitoring Biodiversity. Smithsonian Institution Press, Washington.

Aide, T.M., Zimmerman, J.K., Rosario, M., & Marcano, H. (1996) Forest recovery in abandoned cattle pastures along an elevational gradient in northeastern Puerto Rico. Biotropica, 28, 573-548.

Alonso, L.E. (2000). Ants as indicators of diversity. In Ants: Standard Methods for Measuring and Monitoring Biodiversity (eds D. Agosti, J.D. Majer, L.E. Alonso & T.R. Schultz), pp. 80-88. Smithsonian Institution Press, Washington.

Andersen, A.N. (1986) Diversity, seasonality and community organization of ants at adjacent heath and woodland sites in south-eastern Australia. Australian Journal of Zoology, 34, 53-64.

Andersen, A.N. (1990) The use of ant communities to evaluate changes in Australian terrestrial ecosystems: a review and a recipe. Proceedings of the Ecological Society of Australia, 16, 347-357.

Andersen, A.N. (1991a) Responses of ground-foraging ant communities to three experimental fire regimes in a savanna forest of tropical Australia. Biotropica, 23, 575-585.

Andersen, A.N. (1991b) Sampling communities of ground-foraging ants: pitfall catches compared with quadrat counts in an Australian tropical savanna. Australian Journal of Ecology, 16, 273-279.

Andersen, A.N. (1991c). Parallels between ants and plants: implications for community ecology. In Ant-Plant Interactions (eds C.R. Huxley & D.F. Cutler), pp. 539-558. Oxford University Press, New York.

Andersen, A.N. (1993) Ants as indicators of restoration success at a uranium mine in tropical Australia. Restoration Ecology, 1, 156-167.

Andersen, A.N. (1995a) A classification of Australian ant communities, based on functional groups which parallel plant life-forms in relation to stress and disturbance. Journal of Biogeography, 22, 15-29.

157

Andersen, A.N. (1995b) Measuring more of biodiversity: genus richness as a surrogate for species richness in Australian ant faunas. Biological Conservation, 73, 39-43.

Andersen, A.N. (1997a) Using ants as bioindicators: multiscale issues in ant community ecology. Conservation Ecology [online], 1, 8.

Andersen, A.N. (1997b) Functional groups and patterns of organization in North American ant communities: a comparison with Australia. Journal of Biogeography, 24, 433-460.

Andersen, A.N. (2000a). Global ecology of rainforest ants: functional groups in relation to environmental stress and disturbance. In Ants: Standard Methods for Measuring and Monitoring Biodiversity (eds D. Agosti, J.D. Majer, L.E. Alonso & T.R. Schultz), pp. 25-34. Smithsonian Institution Press, Washington.

Andersen, A.N. (2000b) The Ants of Northern Australia: a Guide to the Monsoonal Fauna. CSIRO Publishing, Collingwood.

Andersen, A.N. & McKaige, M.E. (1987) Ant communities at Rotamah Island, Victoria, with particular reference to disturbance and Rhytidoponera tasmaniensis. Proceedings of the Royal Society of Victoria, 99, 141-146.

Andersen, A.N. & Majer, J.D. (2004) Ants show the way Down Under: invertebrates as bioindicators in land management. Frontiers in Ecology & the Environment, 2, 291-298.

Andersen, A.N., Hoffmann, B.D., & Somes, J. (2003) Ants as indicators of minesite restoration: community recovery at one of eight rehabilitation sites in central Queensland. Ecological Management and Restoration, 4, S12-S19.

Andersen, A.N., Hoffmann, B.D., Muller, W.J., & Griffiths, A.D. (2002) Using ants as bioindicators in land management: simplifying assessment of ant community responses. Journal of Applied Ecology, 39, 8-17.

Andersen, A.N., Fisher, A., Hoffmann, B.D., Read, J.L., & Richards, B. (2004) Use of terrestrial invertebrates for biodiversity monitoring in Australian rangelands, with particular reference to ants. Austral Ecology, 29, 87-92.

Anderson, J.M. (1977) The organization of soil animal communities. Ecological Bulletin, 25, 15-23.

Anderson, M.J. (2005) PERMANOVA: a FORTRAN Computer Program for Permutational Multivariate Analysis of Variance. Department of Statistics, University of Auckland, Auckland.

Andre, H.M., Noti, M.I., & Lebrun, P. (1994) The soil fauna: the other last biotic frontier. Biodiversity and Conservation, 3, 45-56.

Andre, H.M., Ducarme, X., & Lebrun, P. (2002) Soil biodiversity: myth, reality or conning? Oikos, 96, 3-24.

158

Andres, P. & Mateos, E. (2006) Soil mesofaunal responses to post-mining restoration treatments. Applied Soil Ecology, 33, 67-78.

Andrew, N., Rodgerson, L., & York, A. (2000) Frequent fuel-reduction burning: the role of logs and associated leaf litter in the conservation of ant biodiversity. Austral Ecology, 25, 99-107.

Anthony, C.D., Formanowicz, D.R., & Brodie, E.D. (1991) Effect of mine reclamation on ant community structure in eastern Texas. Texas Journal of Science, 43, 261-272.

Aoki, J. & Harada, H. (1982) Environmental change and soil fauna in east Kalimantan (Borneo). Bulletin of the Institute of Environmental Science and Technology, Yokohama National University, 8, 341-378.

Armbrecht, I. & Perfecto, I. (2003) Litter-twig dwelling ant species richness and predation potential within a forest fragment and neighboring coffee plantations of contrasting habitat quality in Mexico. Agriculture Ecosystems & Environment, 97, 107-115.

Armbrecht, I., Perfecto, I., & Vandermeer, J. (2004) Enigmatic biodiversity correlations: ant diversity responds to diverse resources. Science, 304, 284-286.

Armbrecht, I., Rivera, L., & Perfecto, I. (2005) Reduced diversity and complexity in the leaf-litter ant assemblage of Colombian coffee plantations. Conservation Biology, 19, 897-907.

Armstrong, J.K. (1993). Restoration of function or diversity? In Nature Conservation 3: Reconstruction of Fragmented Ecosystems (eds D.A. Saunders, R.J. Hobbs & P.R. Ehrlich), pp. 209-214. Surrey Beatty & Sons, Chipping Norton.

As, S. (1993) Are habitat islands islands? Woodliving beetles (Coleoptera) in deciduous forest fragment in boreal forest. Ecography, 16, 219-228.

Ashton, M.S., Gunatilleke, C.V.S., Singhakumara, B.M.P., & Gunatilleke, I. (2001) Restoration pathways for rain forest in southwest Sri Lanka: a review of concepts and models. Forest Ecology and Management, 154, 409-430.

Asner, G.P., Elmore, A.J., Olander, L.P., Martin, R.E., & Harris, A.T. (2004) Grazing systems, ecosystem responses, and global change. Annual Review of Environment and Resources, 29, 261-299.

Athias-Binche, F. (1993) Dispersal in varying environments - the case of phoretic uropodid mites. Canadian Journal of Zoology, 71, 1793-1798.

Australian State of the Environment Committee (2006) Australia State of the Environment 2006. Department of the Environment and Heritage, Canberra.

Azarbayjani, F.F., Burgin, S., & Richardson, B.J. (1999) Arboreal arthropod biodiversity in woodlands. II. The pattern of recovery of diversity on Melaleuca linariifolia following defaunation. Australian Journal of Ecology, 24, 655-661.

159

Azevedo-Ramos, C., de Carvalho, O., & do Amaral, B.D. (2006) Short-term effects of reduced-impact logging on eastern Amazon fauna. Forest Ecology and Management, 232, 26-35.

Baldi, A. (2003) Using higher taxa as surrogates of species richness: a study based on 3700 Coleoptera, Diptera, and Acari species in CentraI-Hungarian reserves. Basic and Applied Ecology, 4, 589-593.

Balmford, A., Jayasuriya, A.H.M., & Green, M.J.B. (1996) Using higher-taxon richness as a surrogate for species richness: II. Local applications. Proceedings of the Royal Society of London. Series B, 263, 1571-1575.

Basset, Y., Novotny, V., Miller, S., & Springate, N.E. (1998) Assessing the impact of forest disturbance on tropical invertebrates: some comments. Journal of Applied Ecology, 35, 461-466.

Basset, Y., Mavoungou, J.F., Mikissa, J.B., Missa, O., Miller, S.E., Kitching, R.L., & Alonso, A. (2004) Discriminatory power of different arthropod data sets for the biological monitoring of anthropogenic disturbance in tropical forests. Biodiversity and Conservation, 13, 709-732.

Basu, P. (1997) Seasonal and spatial patterns in ground foraging ants in a rain forest in the Western Ghats, India. Biotropica, 29, 489-500.

Beare, M.H., Reddy, M.V., Tian, G., & Srivastava, S.C. (1997) Agricultural intensification, soil biodiversity and agroecosystem function in the tropics: the role of decomposer biota. Applied Soil Ecology, 6, 87-108.

Beattie, A.J. & Oliver, I. (1994) Taxonomic minimalism. Trends in Ecology and Evolution, 9, 488-490.

Beaulieu, F. & Weeks, A.R. (2007) Free-living mesostigmatic mites in Australia: their roles in biological control and bioindication. Australian Journal of Experimental Agriculture, 47, 460-478.

Bengtsson, J. (1998) Which species? What kind of diversity? Which ecosystem function? Some problems in studies of relations between biodiversity and ecosystem function. Applied Soil Ecology, 10, 191-199.

Bennett, A.F., Kimber, S.L., & Ryan, P.A. (2000) Revegetation and Wildlife: a Guide to Enhancing Revegetated Habitats for Wildlife Conservation in Rural Environments. Bushcare National and Research and Development Program Research Report 2/00, Canberra.

Berg, B. & McClaugherty, C. (2003) Plant Litter: Decomposition, Humus Formation, Carbon Sequestration. Springer, Berlin.

Bestelmeyer, B.T. & Wiens, J.A. (1996) The effects of land use on the structure of ground-foraging ant communities in the Argentine Chaco. Ecological Applications, 6, 1225-1240.

160

Bestelmeyer, B.T., Agosti, D., Alonso, L.E., Brandao, C.R.F., Brown, W.L., Delabie, J.H.C., & Silvestre, R. (2000). Field techniques for the study of ground-dwelling ants: an overview, description, and evaluation. In Ants: Standard Methods for Measuring and Monitoring Biodiversity (eds D. Agosti, J.D. Majer, L.E. Alonso & T.R. Schultz), pp. 122-144. Smithsonian Institution Press, Washington.

Big Scrub Rainforest Landcare Group (2005) Subtropical Rainforest Restoration: a Practical Manual and Data Source for Landcare Groups, Land Managers and Rainforest Regenerators, 2nd edn. Big Scrub Rainforest Landcare Group, Bangalow.

Bisevac, L. & Majer, J.D. (1999) Comparative study of ant communities of rehabilitated mineral sand mines and heathland, Western Australia. Restoration Ecology, 7, 117-126.

Bisevac, L. & Majer, J. (2002) Cost effectiveness and data-yield of biodiversity surveys. Journal of the Royal Society of Western Australia, 85, 129-132.

Blake, S., McCracken, D.I., Eyre, M.D., Garside, A., & Foster, G.N. (2003) The relationship between the classification of Scottish ground beetle assemblages (Coleoptera, Carabidae) and the National Vegetation Classification of British plant communities. Ecography, 26, 602-616.

Blanco-Garcia, A. & Lindig-Cisneros, R. (2005) Incorporating restoration in sustainable forestry management: using pine-bark mulch to improve native species establishment on tephra deposits. Restoration Ecology, 13, 703-709.

Block, W.M., Franklin, A.B., Ward, J.P., Ganey, J.L., & White, G.C. (2001) Design and implementation of monitoring studies to evaluate the success of ecological restoration on wildlife. Restoration Ecology, 9, 293-303.

Botes, A., McGeoch, M.A., Robertson, H.G., Niekerk, A., Davids, H.P., & Chown, S.L. (2006) Ants, altitude and change in the northern Cape Floristic Region. Journal of Biogeography, 33, 71-90.

Boulton, A.M. & Amberman, K.D. (2006) How ant nests increase soil biota richness and abundance: a field experiment. Biodiversity and Conservation, 15, 69-82.

Boulton, A.M., Davies, K.F., & Ward, P.S. (2005) Species richness, abundance, and composition of ground-dwelling ants in northern California grasslands: role of plants, soil, and grazing. Environmental Entomology, 34, 96-104.

Bowie, M.H., Marris, J.W.M., Emberson, R.M., Andrew, I.G., Berry, J.A., Vink, C.J., White, E.G., Stufkins, A.W., Oliver, E.H.A., Early, J.W., Klimaszewski, J., Johns, P.M., Wratten, S.D., Mahlfeld, K., Brown, B., Eyles, A.C., Pawson, S.M., & Macfarlane, R.P. (2003) A terrestrial invertebrate inventory of Quail Island (Otamahua): towards the restoration of the invertebrate community. New Zealand Natural Sciences, 28, 81-109.

Bradshaw, A.D. (1983) The reconstruction of ecosystems. Journal of Applied Ecology, 20, 1-17.

161

Bradshaw, A.D. (1984) Ecological principles and land reclamation practice. Landscape and Planning, 11, 35-48.

Bradshaw, A.D. (1987). Restoration: an acid test for ecology. In Restoration Ecology: a Synthetic Approach to Ecological Research (eds W.R.I. Jordan, M.E. Gilpin & J.D. Aber), pp. 23-29. Cambridge University Press, Cambridge.

Bradshaw, A.D. (1996) Underlying principles of restoration. Canadian Journal of Fisheries and Aquatic Sciences, 53, 3-9.

Brady, V.J., Cardinale, B.J., Gathman, J.P., & Burton, T.M. (2002) Does facilitation of faunal recruitment benefit ecosystem restoration? An experimental study of invertebrate assemblages in wetland mesocosms. Restoration Ecology, 10, 617-626.

Brose, U. (2003) Island biogeography of temporary wetland carabid beetle communities. Journal of Biogeography, 30, 879-888.

Brower, A.V.Z. (1995) Taxonomic minimalism. Trends in Ecology & Evolution, 10, 203.

Brown, S. & Lugo, A.E. (1994) Rehabilitation of tropical lands: a key to sustaining development. Restoration Ecology, 2, 97-111.

Brown, S.C., Smith, K., & Batzer, D. (1997) Macroinvertebrate responses to wetland restoration in northern New York. Environmental Entomology, 26, 1016-1024.

Brown, S.E. & Lugo, A.E. (1990) Tropical secondary forests. Journal of Tropical Ecology, 6, 1-32.

Brown, W.L. (2000). Diversity of ants. In Ants: Standard Methods for Measuring and Monitoring Biodiversity (eds D. Agosti, J.D. Majer, L.E. Alonso & T.R. Schultz), pp. 45-79. Smithsonian Institution Press, Washington.

Brust, G.E. (1990) Direct and indirect effects of four herbicides on the activity of carabid beetles (Coleoptera: Carabidae). Pesticide Science, 30, 309-320.

Buckley, R.C. (1982). Ant-plant interactions: a world review. In Ant-Plant Interactions in Australia (ed R.C. Buckley), pp. 111-141. Dr. W. Junk Publishers, The Hague.

Burbidge, A.H., Leicester, K., McDavitt, S., & Majer, J.D. (1992) Ants as indicators of disturbance at Yanchep National Park, Western Australia. Journal of the Royal Society of Western Australia, 75, 89-95.

Butcher, J., Majer, J.D., & Unsworth, P. (1989). Bibliography of fauna studies in reclaimed lands. In Animals in Primary Succession: the Role of Fauna in Reclaimed Lands (ed J.D. Majer), pp. 451-516. Cambridge University Press, Cambridge.

Cabin, R.J. (2007) Science-driven restoration: a square grid on a round earth? Restoration Ecology, 15, 1-7.

Campbell, I.C. (1995) Reply from A.J. Beattie and I. Oliver. Trends in Ecology & Evolution, 10, 203.

162

Cardoso, P., Silva, I., de Oliveira, N.G., & Serrano, A.R.M. (2004) Indicator taxa of spider (Araneae) diversity and their efficiency in conservation. Biological Conservation, 120, 517-524.

Caruso, T. & Migliorini, M. (2006) Micro-arthropod communities under human disturbance: is taxonomic aggregation a valuable tool for detecting multivariate change? Evidence from Mediterranean soil oribatid coenoses. Acta Oecologica, 30, 46-53.

Catterall, C.P. & Harrison, D.A. (2006) Rainforest Restoration Activities in Australia's Tropics and Subtropics. Cooperative Research Centre for Tropical Rainforest Ecology and Management, Cairns.

Catterall, C.P., Kanowski, J., & Wardell-Johnson, G.W. (2007). Biodiversity and new forests: interacting processes, prospects and pitfalls of rainforest restoration. In Living in a Dynamic Tropical Forest Landscape (eds N. Stork & S. Turton). Wiley-Blackwell, Oxford. In Press.

Catterall, C.P., Kanowski, J., Lamb, D., Killin, D., Erskine, P.D., & Wardell-Johnson, G. (2005). Trade-offs between timber production and biodiversity in rainforest plantations: emerging issues and an ecological perspective. In Reforestation in the Tropics and Subtropics of Australia Using Rainforest Tree Species (eds P.D. Erskine, D. Lamb & M. Bristow), pp. 206-220. Rural Industries Research and Development Corporation, Canberra.

Catterall, C.P., Kanowski, J., Wardell-Johnson, G.W., Proctor, H., Reis, T., Harrison, D.A., & Tucker, N.I.J. (2004). Quantifying the biodiversity values of reforestation: perspectives, design issues and outcomes in Australian rainforest landscapes. In Conservation of Australia's Forest Fauna (ed D. Lunney), pp. 359-393. Royal Zoological Society of New South Wales, Mosman.

Chazdon, R.L. (2003) Tropical forest recovery: legacies of human impact and natural disturbances. Perspectives in Plant Ecology, Evolution and Systematics, 6, 51-71.

Chikoski, J.M., Ferguson, S.H., & Meyer, L. (2006) Effects of water addition on soil arthropods and soil characteristics in a precipitation-limited environment. Acta Oecologica, 30, 203-211.

Chung, A.Y.C., Eggleton, P., Speight, M.R., Hammond, P.M., & Chey, V.K. (2000) The diversity of beetle assemblages in different habitat types in Sabah, Malaysia. Bulletin of Entomological Research, 90, 475-496.

Churchill, T.B. (1997) Spiders as ecological indicators: an overview for Australia. Memoirs of the Museum of Victoria, 56, 331-337.

Clapperton, M.J., Kanashiro, D.A., & Behan-Pelletier, V.M. (2002) Changes in abundance and diversity of microarthropods associated with Fescue Prairie grazing regimes. Pedobiologia, 46, 496-511.

Clarke, K.R. (1993) Non-parametric multivariate analyses of changes in community structure. Australian Journal of Ecology, 18, 117-143.

163

Clewell, A. & Rieger, J.P. (1997) What practitioners need from restoration ecologists. Restoration Ecology, 5, 350-354.

Collins, N.M. & Thomas, J.A., eds. (1991) The Conservation of Insects and Their Habitats. Academic Press, London.

Colwell, R.K. (2004) EstimateS: Statistical Estimation of Species Richness and Shared Species from Samples. Version 7. User's Guide and Application. Published at: http://purl.oclc.org/estimates.

Compton, S.G. (2002). Sailing with the wind: dispersal by small flying insects. In Dispersal Ecology (eds J.M. Bullock, R.E. Kenward & R.S. Hails), pp. 113-133. Blackwell, Malden.

Cornish, P.S. & Burgin, S. (2005) Residual effects of glyphosate herbicide in ecological restoration. Restoration Ecology, 13, 695-702.

Cranston, P.S. & Trueman, J.W.H. (1997) "Indicator" taxa in invertebrate biodiversity assessment. Memoirs of the Museum of Victoria, 56, 267-274.

Crisp, P.N., Dickinson, K.J.M., & Gibbs, G.W. (1998) Does native invertebrate diversity reflect native plant diversity? A case study from New Zealand and implications for conservation. Biological Conservation, 83, 209-220.

Cuccovia, A. & Kinnear, A. (1999). Acarine (mite) communities colonizing rehabilitated bauxite mine pits in the jarrah forest of Western Australia. In The Other 99%: the Conservation and Biodiversity of Invertebrates (eds W. Ponder & D. Lunney), pp. 54-59. Royal Zoological Society of New South Wales, Mosman.

Cummings, J., Reid, N., Davies, I., & Grant, C. (2005) Adaptive restoration of sand-mined areas for biological conservation. Journal of Applied Ecology, 42, 160-170.

Curry, J.P. & Good, J.A. (1992) Soil faunal degradation and restoration. Advances in Soil Science, 17, 171-215.

Dangerfield, J.M. (1996) Millipede fecal pellet production in selected natural and managed habitats of southern Africa: implications for litter dynamics. Biotropica, 28, 113-120.

Dauber, J., Purtauf, T., Allspach, A., Frisch, J., Voigtlander, K., & Wolters, V. (2005) Local vs. landscape controls on diversity: a test using surface-dwelling soil macroinvertebrates of differing mobility. Global Ecology and Biogeography, 14, 213-221.

Davies, R.G., Eggleton, P., Dibog, L., Lawton, J.H., Bignell, D.E., Brauman, A., Hartmann, C., Nunes, L., Holt, J., & Rouland, C. (1999) Successional response of a tropical forest termite assemblage to experimental habitat perturbation. Journal of Applied Ecology, 36, 946-962.

164

Davis, A.L.V., van Aarde, R.J., Scholtz, C.H., & Delport, J.H. (2002) Increasing representation of localized dung beetles across a chronosequence of regenerating vegetation and natural dune forest in South Africa. Global Ecology and Biogeography, 11, 191-209.

Davis, A.L.V., van Aarde, R.J., Scholtz, C.H., & Delport, J.H. (2003) Convergence between dung beetle assemblages of a post-mining vegetational chronosequence and unmined dune forest. Restoration Ecology, 11, 29-42.

Davis, B.N.K. (1963) A study of micro-arthropod communities in mineral soils near Corby, Northants. Journal of Animal Ecology, 32, 49-71.

Davis, B.N.K. & Jones, P.E. (1986) Insects on isolated colonies of common rock-rose, Helianthemum chamaecistus. Ecological Entomology, 11, 267-281.

Decaens, T., Dutoit, T., Alard, D., & Lavelle, P. (1998) Factors influencing soil macrofaunal communities in post-pastoral successions of western France. Applied Soil Ecology, 9, 361-367.

Decaens, T., Jimenez, J.J., Gioia, C., Measey, G.J., & Lavelle, P. (2006) The values of soil animals for conservation biology. European Journal of Soil Biology, 42, S23-S38.

Delabie, J.H.C., Fisher, B.L., Majer, J.D., & Wright, I.W. (2000). Sampling effort and choice of methods. In Ants: Standard Methods for Measuring and Monitoring Biodiversity (eds D. Agosti, J.D. Majer, L.E. Alonso & T.R. Schultz), pp. 145-154. Smithsonian Institution Press, Washington. den Boer, P.J. (1990) The survival value of dispersal in terrestrial arthropods. Biological Conservation, 54, 175-192.

Didham, R.K., Ghazoul, J., Stork, N.E., & Davis, A.J. (1996) Insects in fragmented forests: a functional approach. Trends in Ecology and Evolution, 11, 255-260.

Dindal, D.L., ed. (1990) Soil Biology Guide. John Wiley & Sons, New York.

Driscoll, D.A. (2005) Is the matrix a sea? Habitat specificity in a naturally fragmented landscape. Ecological Entomology, 30, 8-16.

Dufrene, M. & Legendre, P. (1997) Species assemblages and indicator species: the need for a flexible asymmetrical approach. Ecological Monographs, 67, 345-366.

Dunger, W. (1989). The return of soil fauna to coal mined areas in the German Democratic Republic. In Animals in Primary Succession: the Role of Fauna in Reclaimed Lands (ed J.D. Majer), pp. 307-337. Cambridge University Press, Cambridge.

Dunger, W., Wanner, M., Hauser, H., Hohberg, K., Schulz, H.-J., Schwalbe, T., Seifert, B., Vogel, J., Voigtlander, K., Zimdars, B., & Zulka, K.P. (2001) Development of soil fauna at mine sites during 46 years after afforestation. Pedobiologia, 45, 243-271.

165

Dunlop, J.N. & Majer, J.D. (1985) A preliminary assessment of minesite rehabilitation in the Pilbara iron ore province using ant communities as ecological indicators. Mulga Research Centre Journal, 8, 25-31.

Eberbach, P.L. & Douglas, L.A. (1983) Persistence of glyphosate in a sandy loam. Soil Biology and Biochemistry, 15, 485-487.

Ehrenfeld, J.G. (2000) Defining the limits of restoration: the need for realistic goals. Restoration Ecology, 8, 2-9.

Eijsackers, H. (1985). Effects of glyphosate on the soil fauna. In The Herbicide Glyphosate (eds E. Grossbard & D. Atkinson), pp. 151-158. Butterworths, London.

Entling, W., Schmidt, M.H., Bacher, S., Brandl, R., & Nentwig, W. (2007) Niche properties of Central European spiders: shading, moisture and the evolution of the habitat niche. Global Ecology and Biogeography, 16, 440-448.

Erskine, P.D. (2002) Land clearing and forest rehabilitation in the Wet Tropics of North Queensland, Australia. Ecological Management and Restoration, 3, 135-137.

Erskine, P.D., Lamb, D., & Bristow, M., eds. (2005) Reforestation in the Tropics and Subtropics of Australia Using Rainforest Tree Species. Rural Industries Research and Development Corporation, Canberra.

FAO (2006) Global Forest Resources Assessment 2005: Progress towards Sustainable Forest Management. Publishing Management Service, FAO, Rome.

Ferrier, S., Gray, M.R., Cassis, G.A., & Wilkie, L. (1999). Spatial turnover in species composition of ground-dwelling arthropods, vertebrates and vascular plants in north-east New South Wales: implications for selection of forest reserves. In The Other 99%: the Conservation and Biodiversity of Invertebrates (eds W. Ponder & D. Lunney), pp. 68-76. Royal Zoological Society of New South Wales, Mosman.

Fimbel, R.A. & Kuser, J.E. (1993) Restoring the pygmy pine forests of New Jersey's Pine Barrens. Restoration Ecology, 1, 117-129.

Finegan, B. (1996) Pattern and process in neotropical secondary rain forests: the first 100 years of succession. Trends in Ecology & Evolution, 11, 119-124.

Fisher, B.L. (1999) Improving inventory efficiency: a case study of leaf-litter ant diversity in Madagascar. Ecological Applications, 9, 714-731.

Fleischner, T.L. (1994) Ecological costs of livestock grazing in Western North America. Conservation Biology, 8, 629-644.

Florentine, S.K. & Westbrooke, M.E. (2004) Evaluation of alternative approaches to rainforest restoration on abandoned pasturelands in tropical north Queensland, Australia. Land Degradation & Development, 15, 1-13.

Floyd, A.G. (1990) Australian Rainforests in New South Wales. Surrey Beatty & Sons, Chipping Norton.

166

Foley, J.A., DeFries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, S.R., Chapin, F.S., Coe, M.T., Daily, G.C., Gibbs, H.K., Helkowski, J.H., Holloway, T., Howard, E.A., Kucharik, C.J., Monfreda, C., Patz, J.A., Prentice, I.C., Ramankutty, N., & Snyder, P.K. (2005) Global consequences of land use. Science, 309, 570-574.

Folgarait, P.J. (1998) Ant biodiversity and its relationship to ecosystem functioning: a review. Biodiversity and Conservation, 7, 1221-1244.

Fox, M.D. & Fox, B.J. (1982). Evidence for interspecific competition influencing ant species diversity in a regenerating heathland. In Ant-Plant Interactions in Australia (ed R.C. Buckley), pp. 99-110. Dr W. Junk Publishers, The Hague.

Franchini, P. & Rockett, C.L. (1996) Oribatid mites as ''indicator'' species for estimating the environmental impact of conventional and conservation tillage practices. Pedobiologia, 40, 217-225.

Franks, N.R. & Bossert, W.H. (1983). The influence of swarm raiding army ants on the patchiness and diversity of a tropical leaf litter ant community. In Tropical Rain Forest: Ecology and Management (eds S.L. Sutton, T.C. Whitmore & A.C. Chadwick), pp. 151-163. Blackwell Scientific Publications, Oxford.

Frith, D. & Frith, C. (1990) Seasonality of litter invertebrate populations in an Australian upland tropical rain forest. Biotropica, 22, 181-190.

Frouz, J., Keplin, B., Pizl, V., Tajovsky, K., Stary, J., Lukesova, A., Novakova, A., Balik, V., Hanel, L., Materna, J., Duker, C., Chalupsky, J., Rusek, J., & Heinkele, T. (2001) Soil biota and upper soil layer development in two contrasting post-mining chronosequences. Ecological Engineering, 17, 275-284.

Gaston, K.J. (2000) Biodiversity: higher taxon richness. Progress in Physical Geography, 24, 117-127.

Gaston, K.J. & May, R.M. (1992) Taxonomy of taxonomists. Nature, 356, 281-282.

Gentry, A.H. & Lopez-Parodi, J. (1980) Deforestation and increased flooding of the upper Amazon. Science, 210, 1354-1356.

Ghilarov, M.S. (1977) Why so many species and so many individuals can coexist in the soil. Ecological Bulletin, 25, 593-597.

Giesy, J.P., Dobson, S., & Solomon, K.R. (2000) Ecotoxicological risk assessment for Roundup herbicide. Reviews of Environmental Contamination and Toxicology, 167, 35-120.

Giller, P.S. (1996) The diversity of soil communities, the 'poor man's tropical rainforest'. Biodiversity and Conservation, 5, 135-168.

Gladstone, W. & Alexander, T. (2005) A test of the higher-taxon approach in the identification of candidate sites for marine reserves. Biodiversity and Conservation, 14, 3151-3168.

167

Glass, R.L. (1987) Adsorption of glyphosate by soils and clay minerals. Journal of Agricultural and Food Chemistry, 35, 497-500.

Godfray, H.C.J. (2002) Challenges for taxonomy. Nature, 417, 17-19.

Goldstein, P.Z. (1997) How many things are there? A reply to Oliver and Beattie, Beattie and Oliver, Oliver and Beattie, and Oliver and Beattie. Conservation Biology, 11, 571-574.

Goldstein, P.Z. (1999) Functional ecosystems and biodiversity buzzwords. Conservation Biology, 13, 247-255.

Gomez, C., Casellas, D., Oliveras, J., & Bas, J.M. (2003) Structure of ground-foraging ant assemblages in relation to land-use change in the northwestern Mediterranean region. Biodiversity and Conservation, 12, 2135-2146.

Goosem, S. & Tucker, N.I.J. (1995) Repairing the Rainforest: Theory and Practice of Rainforest Re-establishment in North Queensland's Wet Tropics. Cassowary Publications/Wet Tropics Management Authority, Cairns.

Gotelli, N.J. & McCabe, D.J. (2002) Species co-occurrence: a meta-analysis of J. M. Diamond's assembly rules model. Ecology, 83, 2091-2096.

Graham, M.H. (2003) Confronting multicollinearity in ecological multiple regression. Ecology, 84, 2809-2815.

Gratton, C. & Denno, R.F. (2005) Restoration of arthropod assemblages in a Spartina salt marsh following removal of the invasive plant Phragmites australis. Restoration Ecology, 13, 358-372.

Grear, J.S. & Schmitz, O.J. (2005) Effects of grouping behavior and predators on the spatial distribution of a forest floor arthropod. Ecology, 86, 960-971.

Green, R.J. & Catterall, C.P. (1998) The effects of forest clearing and regeneration on the fauna of Wivenhoe Park, south-east Queensland. Wildlife Research, 25, 677 - 690.

Greenslade, J. & Greenslade, P. (1984) Invertebrates and environmental assessment. Environment and Planning, 3, 13-15.

Greenslade, P. (2007) The potential of Collembola to act as indicators of landscape stress in Australia. Australian Journal of Experimental Agriculture, 47, 424-434.

Greenslade, P. & Greenslade, P.J.M. (1971) The use of baits and preservatives in pitfall traps. Journal of the Australian Entomological Society, 10, 253-260.

Greenslade, P. & New, T.R. (1991). Australia: conservation of a continental insect fauna. In The Conservation of Insects and Their Habitats (eds N.M. Collins & J.A. Thomas), pp. 33-70. Academic Press, London.

Greenslade, P. & Majer, J.D. (1993) Recolonization by Collembola of rehabilitated bauxite mines in Western Australia. Australian Journal of Ecology, 18, 385-394.

168

Greenslade, P.J.M. (1973) Sampling ants with pitfall traps: digging-in effects. Insectes Sociaux, 20, 343-353.

Greenslade, P.J.M. (1978). Ants. In The Physical and Biological Features of Kunoth Paddock in Central Australia (ed W.A. Low), pp. 109-113. CSIRO Division of Land Resources Management Technical Paper No. 4, Canberra.

Greenslade, P.J.M. & Halliday, R.B. (1983) Colony dispersion and relationships of meat ants Iridomyrmex purpureus and allies in an arid locality in south Australia. Insectes Sociaux, 30, 82-99.

Grimbacher, P.S. & Catterall, C.P. (2007) How much do site age, habitat structure and spatial isolation influence the restoration of rainforest beetle species assemblages? Biological Conservation, 135, 107-118.

Grimbacher, P.S., Catterall, C.P., & Kitching, R.L. (2006) Beetle species' responses suggest that microclimate mediates fragmentation effects in tropical Australian rainforest. Austral Ecology, 31, 458-470.

Grimbacher, P.S., Catterall, C.P., Kanowski, J., & Proctor, H.C. (2007) Responses of ground-active beetle assemblages to different styles of reforestation on cleared rainforest land. Biodiversity and Conservation, 16, 2167-2184.

Grossbard, E. & Atkinson, D., eds. (1985) The Herbicide Glyphosate. Butterworths, London.

Grove, S.J. (2000) Importance of mature timber habitat in forest management and restoration: what can insects tell us? Ecological Management and Restoration, 1, 62-64.

Guariguata, M.R. & Ostertag, R. (2001) Neotropical secondary forest succession: changes in structural and functional characteristics. Forest Ecology and Management, 148, 185-206.

Gubby, C. (1994) Campbellville and Cedar Days: a Compiled History of the Former South-East Queensland Sawmilling Township of Campbellville and Pioneers of the Associated Timber Industry during the Latter Part of the 19th Century. Department of Primary Industries, The State of Queensland.

Gullison, R.E. (2003) Does forest certification conserve biodiversity? Oryx, 37, 153-165.

Hance, R.J. (1976) Adsorption of glyphosate by soils. Pesticide Science, 7, 363-366.

Hansen, R.A. (1999) Red oak litter promotes a microarthropod functional group that accelerates its decomposition. Plant and Soil, 209, 37-45.

Hansen, R.A. (2000) Effects of habitat complexity and composition on a diverse litter microarthropod assemblage. Ecology, 81, 1120-1132.

Harden, G.J., Fox, M.D., & Fox, B.J. (2004) Monitoring and assessment of restoration of a rainforest remnant at Wingham Brush, NSW. Austral Ecology, 29, 489-507.

169

Harris, J.A., Hobbs, R.J., Higgs, E., & Aronson, J. (2006) Ecological restoration and global climate change. Restoration Ecology, 14, 170-176.

Harrison, S.R., Herbohn, J.L., Tisdell, C.A., & Lamb, D. (2000). Timber production and biodiversity tradeoffs in plantation forestry. In Sustainable Small-scale Forestry: Socio-economic Analysis and Policy (eds S.R. Harrison, J.L. Herbohn & K.F. Herbohn), pp. 65-76. Edward Elgar, Cheltenham.

Hassan, S.A., Bigler, F., Bogenschutz, H., Boller, E., Brun, J., Chiverton, P., Edwards, P., Mansour, F., Naton, E., Ooman, P.A., Polgar, L., Rieckmann, W., Peterson, L.S., Staubli, A., Tavakes, K., Tuset, J.J., Viggiani, G., & Vivas, A.G. (1988) Results of the fourth joint pesticide testing programme carried out by the IOBC/WPRS-working group 'pesticides and beneficial organisms'. Journal of Applied Entomology, 105, 3321-3329.

Hattenschwiler, S., Tiunov, A.V., & Scheu, S. (2005) Biodiversity and litter decomposition in terrestrial ecosystems. Annual Review of Ecology Evolution and Systematics, 36, 191-218.

Haughton, A.J., Wilcox, A., & Chaney, K. (1999a) The effects of different rates of glyphosate on non-target invertebrates in field margins. Aspects of Applied Biology, 54, 185-190.

Haughton, A.J., Bell, J.R., Wilcox, A., & Boatman, N.D. (1999b) The effects of different rates of the herbicide glyphosate on spiders in arable field margins. The Journal of Arachnology, 27, 249-254.

Haughton, A.J., Bell, J.R., Boatman, N.D., & Wilcox, A. (2001a) The effect of the herbicide glyphosate on non-target spiders: Part II. Indirect effects on Lepthyphantes tenuis in field margins. Pest Management Science, 57, 1037-1042.

Haughton, A.J., Bell, J.R., Wilcox, A., & Boatman, N.D. (2001b) The effect of the herbicide glyphosate on non-target spiders: Part I. Direct effects on Lepthyphantes tenuis under laboratory conditions. Pest Management Science, 57, 1033-1036.

Hawkins, B.A. & Cross, E.A. (1982) Patterns of refaunation of reclaimed strip mine spoils by non-terricolous arthropods. Environmental Entomology, 11, 762-775.

Hawkins, C.P. & Macmahon, J.A. (1989) Guilds - the multiple meanings of a concept. Annual Review of Entomology, 34, 423-451.

Heneghan, L., Coleman, D.C., Zou, X., Crossley, D.A., & Haines, B.L. (1999) Soil microarthropod contributions to decomposition dynamics: tropical-temperate comparisons of a single substrate. Ecology, 80, 1873-1882.

Heterick, B. (1997) The interaction between the coastal brown ant, Pheidole megacephala (Fabricius), and other invertebrate fauna of Mt Coot-tha (Brisbane, Australia). Australian Journal of Ecology, 22, 218-221.

Hilty, J. & Merenlender, A. (2000) Faunal indicator taxa selection for monitoring ecosystem health. Biological Conservation, 92, 185-197.

170

Hirst, A.J. (2006) Influence of taxonomic resolution on multivariate analyses of arthropod and macroalgal reef assemblages. Marine Ecology Progress Series, 324, 83-93.

Hobbs, R.J. (1993) Can revegetation assist in the conservation of biodiversity in agricultural areas? Pacific Conservation Biology, 1, 29-38.

Hobbs, R.J. & Hopkins, A.J.M. (1990) From frontier to fragments: European impact on Australia's vegetation. Proceedings of the Ecological Society of Australia, 16, 93-114.

Hobbs, R.J. & Harris, J.A. (2001) Restoration ecology: repairing the earth's ecosystems in the new millennium. Restoration Ecology, 9, 239-246.

Hobbs, R.J., Arico, S., Aronson, J., Baron, J.S., Bridgewater, P., Cramer, V.A., Epstein, P.R., Ewel, J.J., Klink, C.A., Lugo, A.E., Norton, D., Ojima, D., Richardson, D.M., Sanderson, E.W., Valladares, F., Vila, M., Zamora, R., & Zobel, M. (2006) Novel ecosystems: theoretical and management aspects of the new ecological world order. Global Ecology and Biogeography, 15, 1-7.

Hoffmann, B.D. (1998) The big-headed ant Pheidole megacephala: a new threat to monsoonal northwestern Australia. Pacific Conservation Biology, 4, 250-255.

Hoffmann, B.D. (2000) Changes in ant species composition and community organisation along grazing gradients in semi-arid rangelands of the Northern Territory. Rangeland Journal, 22, 171-189.

Hoffmann, B.D. & Andersen, A.N. (2003) Responses of ants to disturbance in Australia, with particular reference to functional groups. Austral Ecology, 28, 444-464.

Hoffmann, B.D., Griffiths, A.D., & Andersen, A.N. (2000) Responses to ant communities to dry sulfur deposition from mining emissions in semi-arid tropical Australia, with implications for the use of functional groups. Austral Ecology, 25, 653-663.

Holland, J.M. & Smith, S. (1999) Sampling epigeal arthropods: an evaluation of fenced pitfall traps using mark-release-recapture and comparisons to unfenced pitfall traps in arable crops. Entomologia Experimentalis et Applicata, 91, 347-357.

Holldobler, B. & Wilson, E.O. (1990) The Ants. Springer-Verlag, Berlin.

Holmes, P.R., Boyce, D.C., & Reed, D.K. (1993) The ground beetle (Coleoptera: Carabidae) fauna of welsh peatland biotopes: factors influencing the distribution of ground beetles and conservation implications. Biological Conservation, 63, 153-161.

Hooper, D.U., Bignell, D.E., Brown, V.K., Brussaard, L., Dangerfield, J.M., Wall, D.H., Wardle, D.A., Coleman, D.C., Giller, K.E., Lavelle, P., Van der Putten, W.H., De Ruiter, P.C., Rusek, J., Silver, W.L., Tiedje, J.M., & Wolters, V. (2000) Interactions between aboveground and belowground biodiversity in terrestrial ecosystems: Patterns, mechanisms, and feedbacks. Bioscience, 50, 1049-1061.

Hopkin, S.P. & Read, H.J. (1992) The Biology of Millipedes. Oxford University Press, Oxford.

171

House, A.P.N. (1997). Ecological process in tropical forest rehabilitation: a perspective. In Proceedings of the 6th International Workshop of BIO-REFOR (eds J. Kikkawa, P. Dart, D. Doley, K. Ishii, D. Lamb & K. Suzuki), pp. 191-196. BIO-REFOR, Brisbane.

House, G.J., Worsham, A.D., Sheets, T.J., & Stinner, R.E. (1987) Herbicide effects on soil arthropod dynamics and wheat straw decomposition in a North Carolina no-tillage agroecosystem. Biology and Fertility of Soils, 4, 109-114.

Hubble, G.D., Isbell, R.F., & Northcote, K.H. (1983). Features of Australian soils. In Soils: an Australian Viewpoint (ed CSIRO Division of Soils), pp. 17-47. CSIRO Publications, Melbourne.

Hughes, J.B., Daily, G.C., & Ehrlich, P.R. (2000) Conservation of insect diversity: a habitat approach. Conservation Biology, 14, 1788-1797.

Humphries, C.J., Williams, P.H., & Vane-Wright, R.I. (1995) Measuring biodiversity value for conservation. Annual Review of Ecology and Systematics, 26, 93-111.

Hunter, M.D. & Price, P.W. (1992) Playing chutes and ladders - heterogeneity and the relative roles of bottom-up and top-down forces in natural communities. Ecology, 73, 724-732.

Hutson, B.R. (1980) Colonization of industrial reclamation sites by Acari, Collembola and other invertebrates. Journal of Applied Ecology, 17, 255-275.

Jackson, G.P. & Fox, B.J. (1996) Comparison of regeneration following burning, clearing or mineral sand mining at Tomago, NSW : II. Succession of ant assemblages in a coastal forest. Australian Journal of Ecology, 21, 200-216.

Jansen, A. (1997) Terrestrial invertebrate community structure as an indicator of the success of a tropical rainforest restoration project. Restoration Ecology, 5, 115-124.

Jansen, A. (2005) Avian use of restoration plantings along a creek linking rainforest patches on the Atherton Tablelands, North Queensland. Restoration Ecology, 13, 275-283.

Jenkins, M. (2003) Prospects for biodiversity. Science, 302, 1175-1177.

Judd, K.W. & Mason, C.F. (1995) Colonization of a restored landfill site by invertebrates, with particular reference to the Coleoptera. Pedobiologia, 39, 116-125.

Kalif, K.A.B., Azevedo-Ramos, C., Moutinho, P., & Malcher, S.A.O. (2001) The effect of logging on the ground-foraging ant community in eastern Amazonia. Studies on Neotropical Fauna and Environment, 36, 215-219.

Kaneko, N. & Salamanca, E.F. (1999) Mixed leaf litter effects on decomposition rates and soil microarthropod communities in an oak-pine stand in Japan. Ecological Research, 14, 131-138.

Kanowski, J., Catterall, C.P., Wardell-Johnson, G.W., Proctor, H., & Reis, T. (2003) Development of forest structure on cleared rainforest land in eastern Australia under different styles of reforestation. Forest Ecology and Management, 183, 265-280.

172

Kanowski, J., Catterall, C.P., Proctor, H., Reis, T., Tucker, N.I.J., & Wardell-Johnson, G. (2005). Biodiversity values of timber plantations and restoration plantings for rainforest fauna in tropical and subtropical Australia. In Reforestation in the Tropics and Subtropics of Australia Using Rainforest Tree Species (eds P.D. Erskine, D. Lamb & M. Bristow), pp. 183-205. Rural Industries Research and Development Corporation, Canberra.

Kanowski, J.J., Reis, T.M., Catterall, C.P., & Piper, S.D. (2006) Factors affecting the use of reforested sites by reptiles in cleared rainforest landscapes in tropical and subtropical Australia. Restoration Ecology, 14, 67-76.

Kareiva, P. (1993) Biodiversity - no shortcuts in new maps. Nature, 365, 292-293.

Karna, J., Hansen, E., & Juslin, H. (2003) Environmental activity and forest certification in marketing of forest products - a case study in Europe. Silva Fennica, 37, 253-267.

Kaspari, M. (1996) Litter ant patchiness at the 1-m2 scale: disturbance dynamics in three Neotropical forests. Oecologia, 107, 265-273.

Kaspari, M. & Weiser, M.D. (1999) The size-grain hypothesis and interspecific scaling in ants. Functional Ecology, 13, 530-538.

Kattan, G.H., Correa, D., Escobar, F., & Medina, C. (2006) Leaf-litter arthropods in restored forests in the Colombian Andes: a comparison between secondary forest and tree plantations. Restoration Ecology, 14, 95-102.

Keals, K. & Majer, J.D. (1991). The conservation status of ant communities along the Wulbin-Perenjori Corridor. In Nature Conservation 2: the Role of Corridors (eds D.A. Saunders & R.J. Hobbs), pp. 387-393. Surrey Beatty & Sons, Chipping Norton.

Keesing, V. & Wratten, S.D. (1998) Indigenous invertebrate components in ecological restoration in agricultural landscapes. New Zealand Journal of Ecology, 22, 99-104.

Kim, K.C. & Byrne, L.B. (2006) Biodiversity loss and the taxonomic bottleneck: emerging biodiversity science. Ecological Research, 21, 794-810.

King, J.R., Andersen, A.N., & Cutter, A.D. (1998) Ants as bioindicators of habitat disturbance: validation of the functional group model for Australia's humid tropics. Biodiversity and Conservation, 7, 1627-1638.

Kinnear, A. (1991) Acarine communities of semi-arid soils from the Eastern Goldfields region of Western Australia. Pedobiologia, 35, 273-283.

Kitching, R.L. (1993a). Towards rapid biodiversity assessment - lessons following studies of arthropods of rainforest canopies. In Rapid Biodiversity Assessment: Proceedings of the Biodiversity Assessment Workshop Macquarie University - 1993, pp. 26-30. Research Unit for Biodiversity and Bioresources, Sydney.

Kitching, R.L. (1993b). Biodiversity and taxonomy: impediment or opportunity? In Conservation Biology in Australia and Oceania (eds C. Moritz & J. Kikkawa), pp. 253-268. Surrey Beatty & Sons, Chipping Norton.

173

Koivula, M., Punttila, P., Haila, Y., & Niemela, J. (1999) Leaf litter and the small-scale distribution of carabid beetles (Coleoptera, Carabidae) in the boreal forest. Ecography, 22, 424-435.

Kooyman, R.M. (1996) Growing Rainforest: Rainforest Restoration and Regeneration: Recommendations for the Humid Sub-tropical Region of Northern New South Wales and South-East Queensland. Greening Australia, Brisbane.

Koponen, S. & Niemela, P. (1995) Ground-living arthropods along pollution gradient in boreal pine forest. Entomologica Fennica, 6, 127-131.

Kotze, D.J. & Samways, M.J. (1999) Support for the multi-taxa approach in biodiversity assessment, as shown by epigaeic invertebrates in an afromontane forest archipelago. Journal of Insect Conservation, 3, 125-143.

Krell, F.T. (2004) Parataxonomy vs. taxonomy in biodiversity studies - pitfalls and applicability of 'morphospecies' sorting. Biodiversity and Conservation, 13, 795-812.

Kremen, C. (2005) Managing ecosystem services: what do we need to know about their ecology? Ecology Letters, 8, 468-479.

Kremen, C., Colwell, R.K., Erwin, T.L., Murphy, D.D., Noss, R.F., & Sanjayan, M.A. (1993) Terrestrial arthropod assemblages: their use in conservation planning. Conservation Biology, 7, 796-808.

Kumssa, D.B., van Aarde, R.J., & Wassenaar, T.D. (2004) The regeneration of soil micro-arthropod assemblages in a rehabilitating coastal dune forest at Richards Bay, South Africa. African Journal of Ecology, 42, 346-354.

Lamb, D., Erskine, P.D., & Parrotta, J.A. (2005) Restoration of degraded tropical forest landscapes. Science, 310, 1628-1632.

Landres, P.B., Verner, J., & Thomas, J.W. (1988) Ecological uses of vertebrate indicator species: a critique. Conservation Biology, 2, 316-328.

Landsberg, J., Morton, S., & James, C. (1999). A comparison of the diversity and indicator potential of arthropods, vertebrates and plants in arid rangelands across Australia. In The Other 99%: the Conservation and Biodiversity of Invertebrates (eds W. Ponder & D. Lunney), pp. 111-120. Royal Zoological Society of New South Wales, Mosman.

Langellotto, G. & Denno, R. (2004) Responses of invertebrate natural enemies to complex-structured habitats: a meta-analytical synthesis. Oecologia, 139, 1-10.

Lassau, S.A. & Hochuli, D.F. (2004) Effects of habitat complexity on ant assemblages. Ecography, 27, 157-164.

Lavelle, P., Decaens, T., Aubert, M., Barot, S., Blouin, M., Bureau, F., Margerie, P., Mora, P., & Rossi, J.-P. (2006) Soil invertebrates and ecosystem services. European Journal of Soil Biology, 42, S3-S15.

174

Lawton, J.H., Bignell, D.E., Bolton, B., Bloemers, G.F., Eggleton, P., Hammond, P.M., Hodda, M., Holt, R.D., Larsen, T.B., Mawdsley, N.A., Stork, N.E., Srivastava, D.S., & Watt, A.D. (1998) Biodiversity inventories, indicator taxa and effects of habitat modification in tropical forest. Nature, 391, 72-76.

Levings, S.C. (1982) Patterns of nest dispersion in a tropical ground ant community. Ecology, 63, 338-344.

Levings, S.C. & Windsor, D.M. (1984) Litter moisture content as a determinant of litter arthropod distribution and abundance during the dry season on Barro Colorado Island, Panama. Biotropica, 16, 125-131.

Lewinsohn, T.M., Freitas, A.V.L., & Prado, P.I. (2005a) Conservation of terrestrial invertebrates and their habitats in Brazil. Conservation Biology, 19, 640-645.

Lewinsohn, T.M., Novotny, V., & Basset, Y. (2005b) Insects on plants: diversity of herbivore assemblages revisited. Annual Review of Ecology Evolution and Systematics, 36, 597-620.

Lindsay, E.A. & French, K. (2004) The impact of the herbicide glyphosate on leaf litter invertebrates within Bitou bush, Chrysanthemoides monilifera ssp rotundata, infestations. Pest Management Science, 60, 1205-1212.

Lobry de Bruyn, L.A. (1993) Ant composition and activity in naturally-vegetated and farmland environments on contrasting soils at Kellerberrin, Western Australia. Soil Biology and Biochemistry, 25, 1043-1056.

Longcore, T. (2003) Terrestrial arthropods as indicators of ecological restoration success in coastal sage scrub (California, U.S.A.). Restoration Ecology, 11, 397-409.

Loreau, M., Naeem, S., Inchausti, P., Bengtsson, J., Grime, J.P., Hector, A., Hooper, D.U., Huston, M.A., Raffaelli, D., Schmid, B., Tilman, D., & Wardle, D.A. (2001) Biodiversity and ecosystem functioning: current knowledge and future challenges. Science, 294, 804-808.

Luff, M.L. (1975) Some features influencing the efficiency of pitfall traps. Oecologia, 19, 345-357.

Lugo, A.E. (1997) The apparent paradox of reestablishing species richness on degraded lands with tree monocultures. Forest Ecology and Management, 99, 9-19.

Lugo, A.E. & Helmer, E. (2004) Emerging forests on abandoned land: Puerto Rico's new forests. Forest Ecology and Management, 190, 145-161.

Lussenhop, J. (1992) Mechanisms of microarthropod-micobial interactions in soil. Advances in Ecological Research, 23, 1-33.

Lyons, J., Trimble, S.W., & Paine, L.K. (2000) Grass versus trees: managing riparian areas to benefit streams of central North America. Journal of the American Water Resources Association, 36, 919-930.

175

MacArthur, R.H. & Wilson, E.O. (1967) The Theory of Island Biogeography. Princeton University Press, Princeton.

Madden, K.E. & Fox, B.J. (1997) Arthropods as indicators of the effects of fluoride pollution on the succession following sand mining. Journal of Applied Ecology, 34, 1239-1256.

Magura, T., Tothmeresz, B., & Elek, Z. (2005) Impacts of leaf-litter addition on carabids in a conifer plantation. Biodiversity and Conservation, 14, 475-491.

Majer, J.D. (1981) The Role of Invertebrates in Bauxite Mine Rehabilitation. Forests Dept. of Western Australia, Perth.

Majer, J.D. (1983) Ants: bio-indicators of minesite rehabilitation, land-use and land conservation. Environmental Management, 7, 375-383.

Majer, J.D. (1985) Recolonization by ants of rehabilitated mineral sand mines on North Stradbroke Island, Queensland, with particular reference to seed removal. Australian Journal of Ecology, 10, 31-48.

Majer, J.D., ed. (1989a) Animals in Primary Succession: the Role of Fauna in Reclaimed Lands. Cambridge University Press, Cambridge.

Majer, J.D. (1989b). Fauna studies and land reclamation technology - a review of the history and need for such studies. In Animals in Primary Succession: the Role of Fauna in Reclaimed Lands (ed J.D. Majer), pp. 5-38. Cambridge University Press, Cambridge.

Majer, J.D. (1990) Rehabilitation of disturbed land: long term prospects for the recolonization of fauna. Proceedings of the Ecological Society of Australia, 16, 509-519.

Majer, J.D. (1992) Ant recolonization of rehabilitated bauxite mines of Pocos-De-Caldas, Brazil. Journal of Tropical Ecology, 8, 97-108.

Majer, J.D. (1996) Ant recolonization of rehabilitated bauxite mines at Trombetas, Para, Brazil. Journal of Tropical Ecology, 12, 257-273.

Majer, J.D. (1997) The use of pitfall traps for sampling ants - a critique. Memoirs of the Museum of Victoria, 56, 323-329.

Majer, J.D. & de Kock, A.E. (1992) Ant recolonization of sand mines near Richard Bay, South Africa: an evaluation of progress with rehabilitation. South African Journal of Science, 88, 31-36.

Majer, J.D. & Nichols, O.G. (1998) Long-term recolonization patterns of ants in Western Australian rehabilitated bauxite mines with reference to their use as indicators of restoration success. Journal of Applied Ecology, 35, 161-182.

Majer, J.D., Sartori, M., Stone, R., & Perriman, W.S. (1982) Recolonization by ants and other invertebrates in rehabilitated mineral sand mines near Eneabba, Western Australia. Reclamation and Revegetation Research, 1, 63-81.

176

Majer, J.D., Day, J.E., Kabay, E.D., & Perriman, W.S. (1984) Recolonization by ants in bauxite mines rehabilitated by a number of different methods. Journal of Applied Ecology, 21, 355-375.

Maleny and District Centenary Committee (1978) By Obi Obi Waters: Stories and Photographs of Early Settlement in the Maleny District, Blackall Range, South Eastern Queensland. Maleny and District Centenary Committee, Nambour.

Mando, A., Brussaard, L., & Stroosnijder, L. (1999) Termite- and mulch-mediated rehabilitation of vegetation on crusted soil in West Africa. Restoration Ecology, 7, 33-41.

Mann, R.M. & Bidwell, J.R. (1999) The toxicity of glyphosate and several glyphosate formulations to four species of southwestern Australian frogs. Archives of Environmental Contamination and Toxicology, 36, 193-199.

Maraun, M., Visser, S., & Scheu, S. (1998) Oribatid mites enhance the recovery of the microbial community after a strong disturbance [freezing and heating in a lab microcosm experiment]. Applied Soil Ecology, 9, 175-181.

Mathieu, J., Rossi, J.P., Grimaldi, M., Mora, P., Lavelle, P., & Rouland, C. (2004) A multi-scale study of soil macrofauna biodiversity in Amazonian pastures. Biology and Fertility of Soils, 40, 300-305.

Matthews, E.G. & Kitching, R.L. (1984) Insect Ecology, 2nd edn. University of Queensland Press, Brisbane.

McGeoch, M.A. (1998) The selection, testing and application of terrestrial insects as bioindicators. Biological Reviews of the Cambridge Philosophical Society, 73, 181-201.

McGeoch, M.A., van Rensburg, B.J., & Botes, A. (2002) The verification and application of bioindicators: a case study of dung beetles in a savanna ecosystem. Journal of Applied Ecology, 39, 661-672.

McGlynn, T.P. (1999) The worldwide transfer of ants: geographical distribution and ecological invasions. Journal of Biogeography, 26, 535-548.

Melbourne, B.A. (1999) Bias in the effect of habitat structure on pitfall traps: an experimental evaluation. Australian Journal of Ecology, 24, 228-239.

Michener, W.K. (1997) Quantitatively evaluating restoration experiments: research design, statistical analysis, and data management considerations. Restoration Ecology, 5, 324-337.

Mohamed, A.I., Nair, G.A., Abbas, H.L., & Kassam, H.H. (1992) Effects of pesticide on the survival, growth and oxygen consumption of Hemilepistus reaumuri (Isopoda Onisicidea). Tropical Zoology, 5, 145-153.

177

Moir, M.L., Brennan, K.E.C., Koch, J.M., Majer, J.D., & Fletcher, M.J. (2005) Restoration of a forest ecosystem: the effects of vegetation and dispersal capabilities on the reassembly of plant-dwelling arthropods. Forest Ecology and Management, 217, 294-306.

Monsanto (2000) Roundup Biactive Herbicide Material Safety Data Sheet. Monsanto Australia Limited, Melbourne.

Mooney, H., Cropper, A., & Reid, W. (2005) Confronting the human dilemma: how can ecosystems provide sustainable services to benefit society? Nature, 434, 561-562.

Moore, J.C., Walter, D.E., & Hunt, H.W. (1988) Arthropod regulation of micro- and mesobiota in below-ground detritus food webs. Annual Review of Entomology, 33, 419-439.

Moran, M.D. (2003) Arguments for rejecting the sequential Bonferroni in ecological studies. Oikos, 100, 403-405.

Moran, V.C. & Southwood, T.R.E. (1982) The guild composition of arthropod communities in trees. Journal of Animal Ecology, 51, 289-306.

Moritz, C., Richardson, K.S., Ferrier, S., Monteith, G.B., Stanisic, J., Williams, S.E., & Whiffin, T. (2001) Biogeographical concordance and efficiency of taxon indicators for establishing conservation priority in a tropical rainforest biota. Proceedings of the Royal Society B: Biological Sciences, 268, 1875-1881.

Morrison, L.W. (2002) Island biogeography and metapopulation dynamics of Bahamian ants. Journal of Biogeography, 29, 387-394.

Mortimer, S.R., Booth, R.G., Harris, S.J., & Brown, V.K. (2002) Effects of initial site management on the Coleoptera assemblages colonising newly established chalk grassland on ex-arable land. Biological Conservation, 104, 301-313.

Myers, N., Mittermeier, R.A., Mittermeier, C.G., Fonseca, G.A.B.d., & Kent, J. (2000) Biodiversity hotspots for conservation priorities. Science, 403, 853-858.

Nakamura, A., Proctor, H., & Catterall, P.C. (2003) Using soil and litter arthropods to assess the state of rainforest restoration. Ecological Management and Restoration, 4, S20-S28.

National Land and Water Resources Audit (2001) Australian Native Vegetation Assessment 2001. Department of the Environment and Heritage, Canberra.

Natural Heritage Trust (2000). National Heritage Trust Community Programs '99-'00. Environment Australia,, Canberra.

Ness, J.H. & Bronstein, I.L. (2004) The effects of invasive ants on prospective ant mutualists. Biological Invasions, 6, 445-461.

Neumann, F.G. (1991) Responses of litter arthropods to major natural or artificial ecological disturbances in mountain ash forest. Australian Journal of Ecology, 16, 19-32.

178

Neumann, U. (1973). Succession of soil fauna in afforested spoil banks of the brown-coal mining district of Cologne. In Ecology and Reclamation of Devastated Land (eds R.J. Hutnik & G. Davis), Vol. 2, pp. 335-348. Gordon and Breach, New York.

New, T.R. (1993) Angels on a pin: dimensions of the crisis in invertebrate conservation. American Zoologist, 33, 623-630.

New, T.R. (1996) Taxonomic focus and quality control in insect surveys for biodiversity conservation. Australian Journal of Entomology, 35, 97-106.

New, T.R. (1998) The role of ground beetles (Coleoptera : Carabidae) in monitoring programmes in Australia. Annales Zoologici Fennici, 35, 163-171.

Nichols, O.G. & Burrows, R. (1985) Recolonization of revegetated bauxite mine sites by predatory invertebrates. Forest Ecology and Management, 10, 49-64.

Nichols, O.G. & Nichols, F.M. (2003) Long-term trends in faunal recolonization after bauxite mining in the Jarrah forest of southwestern Australia. Restoration Ecology, 11, 261-272.

Niemela, J., Haila, Y., & Punttila, P. (1996) The importance of small-scale heterogeneity in boreal forests: variation in diversity in forest-floor invertebrates across the succession gradient. Ecography, 19, 352-368.

Novotny, V. & Basset, Y. (2000) Rare species in communities of tropical insect herbivores: pondering the mystery of singletons. Oikos, 89, 564-572.

O'Hanlon, R.P. & Bolger, T. (1999) The importance of Arcitalitrus dorrieni (Hunt) (Crustacea: Amphipoda: Talitridae) in coniferous litter breakdown. Applied Soil Ecology, 11, 29-33.

Olden, J.D. (2006) Biotic homogenization: a new research agenda for conservation biogeography. Journal of Biogeography, 33, 2027-2039.

Oliver, I. & Beattie, A.J. (1993) A possible method for the rapid assessment of biodiversity. Conservation Biology, 7, 562-568.

Oliver, I. & Beattie, A.J. (1996a) Invertebrate morphospecies as surrogates for species: a case study. Conservation Biology, 10, 99-109.

Oliver, I. & Beattie, A.J. (1996b) Designing a cost-effective invertebrate survey: a test of methods for rapid assessment of biodiversity. Ecological Applications, 6, 594-607.

Ottonetti, L., Tucci, L., & Santini, G. (2006) Recolonization patterns of ants in a rehabilitated lignite mine in central Italy: potential for the use of Mediterranean ants as indicators of restoration processes. Restoration Ecology, 14, 60-66.

Palladini, J.D., Jones, M.G., Sanders, N.J., & Jules, E.S. (2007) The recovery of ant communities in regenerating temperate conifer forests. Forest Ecology and Management, 242, 619-624.

179

Parker, V.T. & Pickett, S.T.A. (1997). Restoration as an ecosystem process: implications of the modern ecological paradigm. In Restoration Ecology and Sustainable Development (eds K.M. Urbanslca, N.R. Webb & P.J. Edwaids), pp. 17-32. Cambridge University Press, Cambridge.

Parmenter, R.R. & Macmahon, J.A. (1987) Early successional patterns of arthropod recolonization on reclaimed strip mines in southwestern Wyoming - the ground-dwelling beetle fauna (Coleoptera). Environmental Entomology, 16, 168-177.

Parmenter, R.R., Parmenter, C.A., & Cheney, C.D. (1989) Factors influencing microhabitat partitioning in arid-land darkling beetles (Tenebrionidae): temperature and water conservation. Journal of Arid Environments, 17, 57-67.

Parmenter, R.R., Macmahon, J.A., & Gilbert, C.A.B. (1991) Early successional patterns of arthropod recolonization on reclaimed Wyoming strip mines - the grasshoppers (Orthoptera, Acrididae) and allied faunas (Orthoptera, Gryllacrididae, Tettigoniidae). Environmental Entomology, 20, 135-142.

Parrotta, J.A., Turnbull, J.W., & Jones, N. (1997) Catalyzing native forest regeneration on degraded tropical lands. Forest Ecology and Management, 99, 1-7.

Passell, H.D. (2000) Recovery of bird species in minimally restored Indonesian tin strip mines. Restoration Ecology, 8, 112-118.

Pearce, J.L. & Venier, L.A. (2006) The use of ground beetles (Coleoptera: Carabidae) and spiders (Araneae) as bioindicators of sustainable forest management: a review. Ecological Indicators, 6, 780-793.

Pearson, D.L. (1994) Selecting indicator taxa for the quantitative assessment of biodiversity. Philosophical Transactions of Royal Society of London Series B, 345, 75-79.

Pearson, D.L. & Lederhouse, R.C. (1987) Thermal ecology and the structure of an assemblage of adult tiger beetle species (Cicindelidae). Oikos, 50, 247-255.

Peck, S.L., McQuaid, B., & Campbell, C.L. (1998) Using ant species (Hymenoptera: Formicidae) as a biological indicator of agroecosystem condition. Environmental Entomology, 27, 1102-1110.

Peeters, C. & Ito, F. (2001) Colony dispersal and the evolution of queen morphology in social hymenoptera. Annual Review of Entomology, 46, 601-630.

Pekar, S. (2002) Differential effects of formaldehyde concentration and detergent on the catching efficiency of surface active arthropods by pitfall traps. Pedobiologia, 46, 539-547.

Perfecto, I. & Vandermeer, J. (2002) Quality of agroecological matrix in a tropical montane landscape: ants in coffee plantations in southern Mexico. Conservation Biology, 16, 174-182.

180

Petal, J. (1978). The role of ants in ecosystems. In Production Ecology of Ants and Termites (ed M.V. Brian), Vol. 13, pp. 293-325. Cambridge University Press, Cambridge.

Petersen, S.L., Roundy, B.A., & Bryant, R.M. (2004) Revegetation methods for high-elevation roadsides at Bryce Canyon National Park, Utah. Restoration Ecology, 12, 248-257.

Pik, A.J., Oliver, I., & Beattie, A.J. (1999) Taxonomic sufficiency in ecological studies of terrestrial invertebrates. Australian Journal of Ecology, 24, 555-562.

Pik, A.J., Dangerfield, J.M., Bramble, R.A., Angus, C., & Nipperess, D.A. (2002) The use of invertebrates to detect small-scale habitat heterogeneity and its application to restoration practices. Environmental Monitoring and Assessment, 75, 179-199.

Pimm, S.L., Ayres, M., Balmford, A., Branch, G., Brandon, K., Brooks, T., Bustamante, R., Costanza, R., Cowling, R., Curran, L.M., Dobson, A., Farber, S., da Fonseca, G.A.B., Gascon, C., Kitching, R., McNeely, J., Lovejoy, T., Mittermeier, R.A., Myers, N., Patz, J.A., Raffle, B., Rapport, D., Raven, P., Roberts, C., Rodriguez, J.P., Rylands, A.B., Tucker, C., Safina, C., Samper, C., Stiassny, M.L.J., Supriatna, J., Hall, D.H., & Wilcove, D. (2001) Can we defy nature's end? Science, 293, 2207-2208.

Prescott, C.E. (2005) Do rates of litter decomposition tell us anything we really need to know? Forest Ecology and Management, 220, 66-74.

Proctor, H.C., Kanowski, J., Wardell-Johnson, G., Reis, T., & Catterall, C.P. (2003) Does diversity beget diversity? A comparison between plant and leaf-litter invertebrate richness from pasture to rainforest. Records of the South Australian Museum Monograph Series, 7, 267-274.

Punttila, P., Haila, Y., Pajunen, T., & Tukia, H. (1991) Colonization of clear-cut forests by ants in the southern Finnish taiga - a quantitative survey. Oikos, 61, 250-262.

Quijon, P.A. & Snelgrove, P.V.R. (2006) The use of coarser taxonomic resolution in studies of predation on marine sedimentary fauna. Journal of Experimental Marine Biology and Ecology, 330, 159-168.

Quinn, G.P. & Keough, M.J. (2002) Experimental Design and Data Analysis for Biologists. Cambridge University Press, Cambridge.

Rainio, J. & Niemela, J. (2003) Ground beetles (Coleoptera: Carabidae) as bioindicators. Biodiversity and Conservation, 12, 487-506.

Read, J.L. (1999) The initial response of a chenopod shrubland plant and invertebrate community to two pulses of intensive cattle grazing. The Rangeland Journal, 21, 169-193.

Reay, S.D. & Norton, D.A. (1999) Assessing the success of restoration plantings in a temperate New Zealand forest. Restoration Ecology, 7, 298-308.

181

Recher, H.F. (1989). Colonization of reclaimed land by animals: an ecologist's overview. In Animals in Primary Succession: the Role of Fauna in Reclaimed Lands (ed J.D. Majer), pp. 441-448. Cambridge University Press, Cambridge.

Redi, B.H., van Aarde, R.J., & Wassenaar, T.D. (2005) Coastal dune forest development and the regeneration of millipede communities. Restoration Ecology, 13, 284-291.

Reichle, D.E. (1977) The role of soil invertebrates in nutrient cycling. Ecological Bulletin, 25, 145-156.

Relyea, R.A. (2005) The lethal impact of Roundup on aquatic and terrestrial amphibians. Ecological Applications, 15, 1118-1124.

Remmert, H. (1981) Body size of terrestrial arthropods and biomass of their populations in relation to the abiotic parameters of their milieu. Oecologia, 50, 12-13.

Rey, J.R. (1981) Ecological biogeography of arthropods on Spartina islands in northwest Florida. Ecological Monographs, 51, 237-265.

Ribas, C.R. & Schoereder, J.H. (2002) Are all ant mosaics caused by competition? Oecologia, 131, 606-611.

Roback, P.J. & Askins, R.A. (2005) Judicious use of multiple hypothesis tests. Conservation Biology, 19, 261-267.

Roth, D.S., Perfecto, I., & Rathcke, B. (1994) The effects of management systems on ground-foraging ant diversity in Costa Rica. Ecological Applications, 4, 423-436.

Ruf, A. (1998) A maturity index for predatory soil mites (Mesostigmata: Gamasina) as an indicator of environmental impacts of pollution on forest soils. Applied Soil Ecology, 9, 447-452.

Ruiz-Jaen, M.C. & Aide, T.M. (2005) Restoration success: how is it being measured? Restoration Ecology, 13, 569-577.

Ruppert, E.E., Fox, R.S., & Barnes, R.D. (2004) Invertebrate Zoology, 7th edn. Thompson Learning, Belmont.

Sakaino, M., Harada, H., & Bae, T.W. (2002) An index of naturalness based on soil animal communities of environmental protection forests. ECO-HABITAT, 9, 119-127.

Sala, O.E., Chapin, F.S., Armesto, J.J., Berlow, E., Bloomfield, J., Dirzo, R., Huber-Sanwald, E., Huenneke, L.F., Jackson, R.B., Kinzig, A., Leemans, R., Lodge, D.M., Mooney, H.A., Oesterheld, M., Poff, N.L., Sykes, M.T., Walker, B.H., Walker, M., & Wall, D.H. (2000) Global biodiversity scenarios for the year 2100. Science, 287, 1770-1774.

Samyn, Y. & Massin, C. (2002) Taxonomists' requiem? Science, 295, 276-277.

Sanchez, B.C. & Parmenter, R.R. (2002) Patterns of shrub-dwelling arthropod diversity across a desert shrubland-grassland ecotone: a test of island biogeographic theory. Journal of Arid Environments, 50, 247-265.

182

Sarty, M., Abbott, K.L., & Lester, P.J. (2006) Habitat complexity facilitates coexistence in a tropical ant community. Oecologia, 149, 465-473.

Schnell, M.R., Pik, A.J., & Dangerfield, J.M. (2003) Ant community succession within eucalypt plantations on used pasture and implications for taxonomic sufficiency in biomonitoring. Austral Ecology, 28, 553-565.

Schulze, C.H., Waltert, M., Kessler, P.J.A., Pitopang, R., Shahabuddin, Veddeler, D., Muhlenberg, M., Gradstein, S.R., Leuschner, C., Steffan-Dewenter, I., & Tscharntke, T. (2004) Biodiversity indicator groups of tropical land-use systems: comparing plants, birds, and insects. Ecological Applications, 14, 1321-1333.

Scott, T.A., Wehtje, W., & Wehtje, M. (2001) The need for strategic planning in passive restoration of wildlife populations. Restoration Ecology, 9, 262-271.

Seastedt, T.R. & Crossley, D.A. (1980) Effects of microarthropods on the seasonal dynamics of nutrients in forest litter. Soil Biology and Biochemistry, 12, 337-342.

Seastedt, T.R. & Crossley, D.A. (1984) The influence of arthropods on ecosystems. Bioscience, 34, 157-161.

SERI (2004) The SER International Primer on Ecological Restoration. Society for Ecological Restoration International, Tucson.

Shattuck, S.O. (1999) Australian Ants: their Biology and Identification. CSIRO Publishing, Collingwood.

Shure, D.J. & Phillips, D.L. (1991) Patch size of forest openings and arthropod populations. Oecologia, 86, 325-334.

Simberloff, D. & Dayan, T. (1991) The guild concept and the structure of ecological communities. Annual Review of Ecology and Systematics, 22, 115-143.

Simberloff, D.S. & Wilson, E.O. (1969) Experimental zoogeography of islands: the colonization of empty islands. Ecology, 50, 278-296.

Simmonds, S.J., Majer, J.D., & Nichols, O.G. (1994) A comparative study of spider (Araneae) communities of rehabilitated bauxite mines and surrounding forest in the southwest of Western Australia. Restoration Ecology, 2, 247-260.

Smit, A.M. & van Aarde, R.J. (2001) The influence of millipedes on selected soil elements: a microcosm study on three species occurring on coastal sand dunes. Functional Ecology, 15, 51-59.

Smith, C.C. (1940) The effect of overgrazing and erosion upon the biota of the mixed-grass prairie of Oklahoma. Ecology, 21, 381-397.

Snedecor, G.W. & Cochran, W.G. (1980) Statistical Methods, 7th edn. The Iowa State University Press, Ames.

Soares, S.M. & Schoereder, J.H. (2001) Ant-nest distribution in a remnant of tropical rainforest in southeastern Brazil. Insectes Sociaux, 48, 280-286.

183

Sodhi, N.S., Koh, L.P., Brook, B.W., & Ng, P.K.L. (2004) Southeast Asian biodiversity: an impending disaster. Trends in Ecology and Evolution, 19, 654-660.

Sousa, J.P. & da Gama, M.M. (1994) Rupture in a Collembola community structure from Quercus rotundifolia Lam forest due to the reafforestation with Eucalyptus globulus Labill. European Journal of Soil Biology, 30, 71-78.

Sprankle, P., Meggitt, W.F., & Penner, D. (1975a) Rapid inactivation of glyphosate in soil. Weed Science, 23, 224-228.

Sprankle, P., Meggitt, W.F., & Penner, D. (1975b) Adsorption, mobility and microbial degradation of glyphosate in the soil. Weed Science, 23, 229-234.

St. John, M.G., Bagatto, G., Behan-Pelletier, V., Lindquist, E.E., Shorthouse, J.D., & Smith, I.M. (2002) Mite (Acari) colonization of vegetated mine tailings near Sudbury, Ontario, Canada. Plant and Soil, 245, 295-305.

Standen, V. (1978) The influence of soil fauna on decomposition by micro-organisms in blanket bog litter. Journal of Animal Ecology, 47, 25-38.

Stanisic, J., Burwell, C., Raven, R., Monteith, G., & Baehr, B. (2005) Terrestrial Invertebrate Status Review Brisbane City. Queensland Museum, Brisbane.

Stephens, S.S. & Wagner, M.R. (2006) Using ground foraging ant (Hymenoptera : Formicidae) functional groups as bioindicators of forest health in northern Arizona ponderosa pine forests. Environmental Entomology, 35, 937-949.

Stork, N.E. (1993) How many species are there? Biodiversity and Conservation, 2, 215-232.

Suding, K.N., Gross, K.L., & Houseman, G.R. (2004) Alternative states and positive feedbacks in restoration ecology. Trends in Ecology and Evolution, 19, 46-53.

Sullivan, T.P. (1985). Effects of glyphosate on selected species of wildlife. In The Herbicide Glyphosate (eds E. Grossbard & D. Atkinson), pp. 186-199. Butterworths, London.

Swart, J.A.A., van der Windt, H.J., & Keulartz, J. (2001) Valuation of nature in conservation and restoration. Restoration Ecology, 9, 230-238.

Swift, M.J., Heal, O.W., & Anderson, J.M. (1979) Decomposition in Terrestrial Ecosystems (Studies in Ecology; vol. 5). University of California Press, California.

Taylor, R.L., Maxwell, B.D., & Boik, R.J. (2006) Indirect effects of herbicides on bird food resources and beneficial arthropods. Agriculture, Ecosystems & Environment, 116, 157-164.

Terborg, J. & Robinson, S. (1986). Guilds and their utility in ecology. In Community Ecology: Pattern and Process (eds J. Kikkawa & D.J. Anderson), pp. 65-90. Blackwell Scientific, Melbourne.

184

Tews, J., Brose, U., Grimm, V., Tielborger, K., Wichmann, M.C., Schwager, M., & Jeltsch, F. (2004) Animal species diversity driven by habitat heterogeneity/diversity: the importance of keystone structures. Journal of Biogeography, 31, 79-92.

Thomas, J.A. (1991). Rare species conservation: case studies of European butterflies. In The Scientific Management of Temperate Communities for Conservation: the 31st Symposium of the British Ecological Society, Southampton, 1989 (eds I.F. Spellerberg, F.B. Goldsmith & M.G. Morris), pp. 149-197. Blackwell Scientific Publications, Oxford.

Tian, G., Brussaard, L., & Kang, B.T. (1993) Biological effects of plant residues with contrasting chemical compositions under humid tropical conditions: effects on soil fauna. Soil Biology and Biochemistry, 25, 731-737.

Tilman, D., Reich, P.B., & Knops, J.M.H. (2006) Biodiversity and ecosystem stability in a decade-long grassland experiment. Nature, 441, 629-632.

Torres, J.A. (1984) Diversity and distribution of ant communities in Puerto Rico. Biotropica, 16, 296-303.

Torstensson, L. (1985). Behaviour of glyphosate in soils and its degradation. In The Herbicide Glyphosate (eds E. Grossbard & D. Atkinson), pp. 137-150. Butterworths, London.

Trueman, J.W.H. & Cranston, P.S. (1997) Prospects for rapid assessment of terrestrial invertebrate biodiversity. Memoirs of the Museum of Victoria, 56, 349-354.

Tsui, M.T.K. & Chu, L.M. (2004) Comparative toxicity of glyphosate-based herbicides: aqueous and sediment porewater exposures. Archives of Environmental Contamination and Toxicology, 46, 316-323.

Tucker, N.I.J. (2001). Wildlife colonisation on restored tropical lands: what can it do, how can we hasten it, and what can we expect? In Forest Restoration for Wildlife Conservation (eds S. Elliott, J. Kerby, D. Blakesley, K. Hardwicke, K. Woods & V. Anusarnsunthorn), pp. 279-295. International Tropical Timber Organisation and the Forest Restoration Research Unit. Chiang Mai University, Thailand.

Tucker, N.I.J., Wardell-Johnson, G., Catterall, C.P., & Kanowski, J. (2004). Agroforestry and biodiversity: improving the outcomes for conservation in tropical north-eastern Australia. In Agroforestry and Biodiversity Conservation in Tropical Landscapes (eds G. Schroth, G. Fonseca, C.A. Harvey, C. Cascon, H. Vasconcelos & A.M.N. Izac), pp. 431-452. Island Press, Washington.

Uetz, G.W. (1979) The influence of variation in litter habitats on spider communities. Oecologia, 40, 29-42.

Uetz, G.W., Halaj, J., & Cady, A.B. (1999) Guild structure of spiders in major crops. Journal of Arachnology, 27, 270-280.

Underwood, E.C. & Fisher, B.L. (2006) The role of ants in conservation monitoring: if, when, and how. Biological Conservation, 132, 166-182.

185

United Nations Department of Economic and Social Affairs (2002) Global Challenge Global Opportunity: Trends in Sustainable Development. The United Nations Department of Economic and Social Affairs for the World Summit on Sustainable Development, Johannesburg. van Aarde, R.J., Ferreira, S.M., & Kritzinger, J.J. (1996a) Millipede communities in rehabilitating coastal dune forests in northern KwaZulu Natal, South Africa. Journal of Zoology, London, 238, 703-712. van Aarde, R.J., Smit, A.-M., & Claassens, A.S. (1998) Soil characteristics of rehabilitating and unmined coastal dunes at Richards Bay, KwaZulu-Natal, South Africa. Restoration Ecology, 6, 102-110. van Aarde, R.J., Ferreira, S.M., Kritzinger, J.J., van Dyk, P.J., Vogt, M., & Wassenaar, T.D. (1996b) An evaluation of habitat rehabilitation on coastal dune forests in northern KwaZulu-Natal, South Africa. Restoration Ecology, 4, 334-345.

Van Dijk, T.S. (1986) Changes in the carabid fauna of a previously agricultural field during the first twelve years of impoverishing treatments. Netherlands Journal of Zoology, 36, 413-437. van Hamburg, H., Andersen, A.N., Meyer, W.J., & Robertson, H.G. (2004) Ant community development on rehabilitated ash dams in the South African highveld. Restoration Ecology, 12, 552-558. van Straalen, N.M. (1997). Community structure of soil arthropods as a bioindicator of soil health. In Biological Indicators of Soil Health (eds C. Pankhurst, B.M. Doube & V.V.S.R. Gupta), pp. 235-264. CAB International, Wallingford. van Straalen, N.M. (1998) Evaluation of bioindicator systems derived from soil arthropod communities. Applied Soil Ecology, 9, 429-437.

Vanderwoude, C., Andersen, A.N., & House, A.P.N. (1997) Community organisation, biogeography and seasonality of ants in an open forest of south-eastern Queensland. Australian Journal of Zoology, 45, 523-537.

Vanderwoude, C., Lobry de Bruyn, L.A., & House, A.P.N. (2000a) Long-term ant community responses to selective harvesting of timber from Spotted Gum (Corymbia variegata)-dominated forests in south-east Queensland. Ecological Management and Restoration, 1, 204-214.

Vanderwoude, C., Lobry de Bruyn, L.A., & House, A.P.N. (2000b) Response of an open-forest ant community to invasion by the introduced ant, Pheidole megacephala. Austral Ecology, 25, 253-259.

Vasconcelos, H.L., Vilhena, J.M.S., & Caliri, G.J.A. (2000) Responses of ants to selective logging of a central Amazonian forest. Journal of Applied Ecology, 37, 508-514.

Veech, J.A., Crist, T.O., & Summerville, K.S. (2003) Intraspecific aggregation decreases local species diversity of arthropods. Ecology, 84, 3376-3383.

186

Villasenor, J.L., Ibarra-Manriquez, G., Meave, J.A., & Ortiz, E. (2005) Higher taxa as surrogates of plant biodiversity in a megadiverse country. Conservation Biology, 19, 232-238.

Walker, B.H. (1992) Biodiversity and ecological redundancy. Conservation Biology, 6, 18-23.

Walter, D.E. & Proctor, H.C. (1999) Mites: Eology, Evolution, and Behaviour. University of New South Wales Press, Sydney.

Wanner, M. & Dunger, W. (2002) Primary immigration and succession of soil organisms on reclaimed opencast coal mining areas in eastern Germany. European Journal of Soil Biology, 38, 137-143.

Ward, D.F. & Stanley, M.C. (2004) The value of RTUs and parataxonomy versus taxonomic species. New Zealand Entomologist, 27, 3-9.

Ward, D.F. & Lariviere, M.C. (2004) Terrestrial invertebrate surveys and rapid biodiversity assessment in New Zealand: lessons from Australia. New Zealand Journal of Ecology, 28, 151-159.

Wardle, D.A. (2006) The influence of biotic interactions on soil biodiversity. Ecology Letters, 9, 870-886.

Wardle, D.A., Nicholson, K.S., Bonner, K.I., & Yeates, G.W. (1999a) Effects of agricultural intensification on soil associated arthropod population dynamics, community structure, diversity and temporal variability over a seven-year period. Soil Biology and Biochemistry, 31, 1691-1706.

Wardle, D.A., Yeates, G.W., Barker, G.M., & Bonner, K.I. (2006) The influence of plant litter diversity on decomposer abundance and diversity. Soil Biology and Biochemistry, 38, 1052-1062.

Wardle, D.A., Bonner, K.I., Barker, G.M., Yeates, G.W., Nicholson, K.S., Bardgett, R.D., Watson, R.N., & Ghani, A. (1999b) Plant removals in perennial grassland: vegetation dynamics, decomposers, soil biodiversity, and ecosystem properties. Ecological Monographs, 69, 535-568.

Wassenaar, T.D., van Aarde, R.J., Pimm, S.L., & Ferreira, S.M. (2005) Community convergence in disturbed subtropical dune forests. Ecology, 86, 655-666.

Watt, A.D., Stork, N.E., Eggleton, P., Diane, S., Bolton, B., Larsen, T., B., Brendell, M.J.D., & Bignell, D.E. (1997). Impact of forest loss and regeneration on insect abundance and diversity. In Forests and Insects (eds A.D. Watt, N.E. Stork & M.D. Hunter), pp. 273-286. Chapman & Hall, London.

Watts, C.H. & Gibbs, G.W. (2002) Revegetation and its effect on the ground-dwelling beetle fauna of Matiu-Somes Island, New Zealand. Restoration Ecology, 10, 96-106.

Webb, C.E., Oliver, I., & Pik, A.J. (2000) Does coastal foredune stabilization with Ammophila arenaria restore plant and arthropod communities in southeastern Australia? Restoration Ecology, 8, 283-288.

187

Weiher, E. (2007) On the status of restoration science: obstacles and opportunities. Restoration Ecology, 15, 340-343.

Wheater, C.P., Cullen, W.R., & Bell, J.R. (2000) Spider communities as tools in monitoring reclaimed limestone quarry landforms. Landscape Ecology, 15, 401-406.

Wheeler, Q.D., Raven, P.H., & Wilson, E.O. (2004) Taxonomy: impediment or expedient? Science, 303, 285-285.

Whisenant, S.G. (1999) Repairing Damaged Wildlands: a Process-Oriented, Landscape-Scale Approach. Cambridge University Press, Cambridge.

White, P.S. & Walker, J.L. (1997) Approximating nature's variation: selecting and using reference information in restoration ecology. Restoration Ecology, 5, 338-349.

Willett, T.R. (2001) Spiders and other arthropods as indicators in old-growth versus logged redwood stands. Restoration Ecology, 9, 410-420.

Williams, K.S. (1993) Use of terrestrial arthropods to evaluate restored riparian woodlands. Restoration Ecology, 1, 107-116.

Williams, K.S. (1997). Terrestrial arthropods as ecological indicators of habitat restoration in southwestern North America. In Restoration Ecology and Sustainable Development (eds K.M. Urbansca, N.R. Webb & P.J. Edwards), pp. 238-258. Cambridge University Press, Cambridge.

Williams, P.H. & Gaston, K.J. (1994) Measuring more of biodiversity: can higher-taxon richness predict wholesale species richness? Biological Conservation, 67, 211-217.

Wilson, A.D. (1990) The effect of grazing on Australian ecosystems. Proceedings of the Ecological Society of Australia, 16, 235-244.

Wilson, E.O. (1958) Patchy distribution of ant species in New Guinea rain forests. Psyche, 65, 26-38.

Wilson, E.O. (1987) The little things that run the world (the importance and conservation of invertebrates). Conservation Biology, 1, 344-346.

Wilson, E.O. (1992) The Diversity of Life. Harvard University Press, Harvard.

Wilson, E.O. & Hunt, G.L. (1966) Habitat selection by the queens of two field-dwelling species of ants. Ecology, 47, 485-487.

Wilson, M.R. (2000) Loss of taxonomists is a threat to pest control. Nature, 407, 559-559.

Woinarski, J.C.Z., Andersen, A.N., Churchill, T.B., & Ash, A.J. (2002) Response of ant and terrestrial spider assemblages to pastoral and military land use, and to landscape position, in a tropical savanna woodland in northern Australia. Austral Ecology, 27, 324-333.

188

Woodford, B.R. (2000) Converting a dairy farm back to a rainforest water catchment. The Rocky Creek Dam story. Ecological Management and Restoration, 1, 83-92.

Yates, C.J. & Hobbs, R.J. (1997) Temperate eucalypt woodlands: a review of their status, processes threatening their persistence and techniques for restoration. Australian Journal of Botany, 45, 949-973.

York, A. (2000) Long-term effects of frequent low-intensity burning on ant communities in coastal blackbutt forests of southeastern Australia. Austral Ecology, 25, 83-98.

Young, T.P. (2000) Restoration ecology and conservation biology. Biological Conservation, 92, 73-83.

Young, T.P., Petersen, D.A., & Clary, J.J. (2005) The ecology of restoration: historical links, emerging issues and unexplored realms. Ecology Letters, 8, 662-673.

189

190

APPENDIX 1

Summary of the previous studies of actively-restored terrestrial habitats which incorporated soil and litter arthropods

191 Appendix 1 Summary of the studies of actively-restored terrestrial habitats which incorporated soil and litter arthropods, with particular attention to the type of restoration, study design and effects of factors considered important for the development of soil and litter arthropod assemblages. Groups of Sampling methods Age of No of arthropods 4 5 6 studied restoration ref Factors investigated

1 2 3 and/visual survey thers aiting Winkler sacs Tullgren/Berlese ext Chronosequence Longitudinal Undisturbed Disturbed

PreUse Habitat Ctry Ants Beetles Spiders arthropods Other traps Pitfall H B O age Restoration techniques Restoration diversity taxonomic Plant (%) Canopy cover by vegetation (%) cover Ground diversity structural Plant height Tree density Tree depth/amount Litter (%) by litter covered Ground debris woody Coarse composition Litter pH) P, N, (C, chemistry Soil penetrability Soil Soil structure content moisture Soil rocks of Presence interactions Inter-specific temperature Ground habitats undisturbed from Distances temperature) (rainfall, Climate Pollution Authors Mine ?TemFor USA XXXX X 0-5yrs - - nn pHawkins & Cross 1982 Mine Heath, AUS X X 4-11yrs 4 - yynn y yn Fox & Fox 1982 Shrub Mine DuneFor ZAF X X 1-21yrs 3 - pyn y Davis et al. 2003; Davis et al. , 2002 Mine DuneFor ZAF X XXXX 0-13yrs 3 - ypnnnn n p Majer & deKock 1992 Mine DuneFor ZAF X X X X 1-24yrs 1 - nyKumssa et al. 2004 Mine DuneFor ZAF X X 2-14yrs 1 - p van Aarde et al. 1996a; van Aarde et al. , 1996b Mine DuneFor ZAF X X 2-24yrs 3 - ypRedi et al. 2005 Mine Grass GBR ?X ?X ?X X X 10-11yrs 1 - pn ppn Davis 1963 Mine Grass USA X X X 1-10yrs 1 - ppp n n n n pn Anthony et al. 1991 Mine Grass USA X X X 0-6yrs 1 - py n n Parmenter et al. 1991 Mine Grass USA X X 0-6yrs 1 - yyy n n Parmenter & Macmahon 1987 Mine Heath AUS X ?X ?X X X X X 1-3yrs 4 - yy npy Majer et al. 1982 Mine RainFor BRA X X X X X X 0-11yrs 3 - nn pp p pnp p Majer 1996 Mine RainFor, BRA X X X X 1-10yrs 2 - ppnpppn nnn n p Majer 1992 Shrub Groups of Sampling methods Age of No of arthropods 4 5 6 studied restoration ref Factors investigated thers 1 2 3 survey and/visual aiting Winkler sacs Tullgren/Berlese ext Chronosequence Longitudinal Undisturbed Disturbed B O age Restoration techniques Restoration diversity taxonomic Plant (%) Canopy cover by vegetation (%) cover Ground diversity structural Plant height Tree density Tree depth/amount Litter (%) by litter covered Ground debris woody Coarse composition Litter pH) P, N, (C, chemistry Soil penetrability Soil structure Soil content moisture Soil rocks of Presence interactions Inter-specific temperature Ground habitats undisturbed from Distances temperature) (rainfall, Climate Pollution PreUse Habitat Ctry Ants Beetles Spiders arthropods Other traps Pitfall H Authors Mine SclFor AUS X X 8yrs 2 - yp p Nichols & Nichols 2003 Mine SclFor AUS X X 14yrs 1 - yyypn n p Majer & Nichols 1998 Mine SclFor AUS XXXX X 2yrs 1 - py pp n Majer 1981 Mine SclFor AUS XXXX X X X 5-13yrs 2 - ny p p pp n Nichols & Burrows 1985 Mine SclFor AUS XXXX X 3-20yrs - - y yyy yyn yMadden & Fox 1997 Mine SclFor AUS X X 1-18yrs - - yy p Jackson & Fox 1996 Mine SclFor AUS X X X X X 2-18yrs 10yrs 2 - ppy yynnn Simmonds et al. 1994 Mine SclFor AUS X X 2-20yrs 1 - pp ppp Cuccovia & Kinnear 1999 Mine SclFor AUS X X X X 0-13yrs 3 1 ypy y nyy n n n Majer et al. 1984 Mine SclFor AUS X X X 2-8yrs 2 - ppnny yyAndersen 1993 Mine SclFor AUS X X 2-10yrs 3 - ypn pAndersen et al. 2003 Mine SclFor AUS X X X X 0-13yrs 3 1 ypyyyyynnpn Greenslade & Majer 1993 Mine SclFor, AUS X X X X X 0-15yrs 3 - ypyy yypn yn Majer 1985 Shrub Mine SclHth AUS X X X X X 2-20yrs 17yrs 4 - yyp n p p Bisevac & Majer 1999 Mine Shrub AUS X X X 2-5yrs 1 1 pp p Dunlop & Majer 1985 Mine TemFor DEU XXXX X X X 37yrs - - ppnn Dunger et al. 2001 Mine TemFor DEU X X X 2-28yrs 2 - pp p p pp Neumann 1973 Groups of Sampling methods Age of No of arthropods 4 5 6 studied restoration ref Factors investigated thers 1 2 3 survey and/visual aiting Winkler sacs Tullgren/Berlese ext Chronosequence Longitudinal Undisturbed Disturbed B O age Restoration techniques Restoration diversity taxonomic Plant (%) Canopy cover by vegetation (%) cover Ground diversity structural Plant height Tree density Tree depth/amount Litter (%) by litter covered Ground debris woody Coarse composition Litter pH) P, N, (C, chemistry Soil penetrability Soil structure Soil content moisture Soil rocks of Presence interactions Inter-specific temperature Ground habitats undisturbed from Distances temperature) (rainfall, Climate Pollution PreUse Habitat Ctry Ants Beetles Spiders arthropods Other traps Pitfall H Authors Mine TemFor DEU X X 1-63yrs 2 - pp p p p Frouz et al. 2001 Mine TemFor DEU XXXX X X X ca.46yrs* - - pyWanner & Dunger 2002 Mine TemFor DEU X X X X X 33yrs - - pp p pDunger 1989 Mine TemFor ESP XXXX X 8-13yrs 2 - yyyAndres & Mateos 2006 Mine TemFor GBR X X X X 0-2yrs 2yrs - - yy p Hutson 1980 Mine TemFor ITA X X 10-20yrs 1 1 ppp Ottonetti et al. 2006 Mine TemFor, CDN X X 0-40yrs 1 1 ppStJohn et al. 2002 Shrub Agr Heath NLD X X 11yrs 1 - p Van Dijk 1986 Agr MontFor COL XXXX X ca.40yrs* 2 - pp pKattan et al. 2006 Agr RainFor AUS X X 5-70yrs 10 5 pyyyyy y n Grimbacher et al. 2007 Agr RainFor AUS X X X X 0-1yr 2 1 pnn pKing et al. 1998 Agr SclFor AUS X X 4, 10yrs 1 1 ypp Schnell et al. 2003 Agr TemFor, NZL X X 5-100yrs 1 - ppppp Watts & Gibbs 2002 Shrub ?Agr TemFor NZL X X X 12-35yrs 1 1 yy py n Reay & Norton 1999 Agr, Clr Shrub USA XXXX X 1-14yrs 7 - nn yy y Longcore 2003 Agr, Weed RainFor AUS X X X 1-12yrs 5 5 pyyyn y Nakamura et al. 2003

Clr RipFor USA XXXX X X 3yrs 1 - ppp Williams 1993 Groups of Sampling methods Age of No of arthropods 4 5 6 studied restoration ref Factors investigated thers 1 2 3 survey and/visual aiting Winkler sacs Tullgren/Berlese ext Chronosequence Longitudinal Undisturbed Disturbed B O age Restoration techniques Restoration diversity taxonomic Plant (%) Canopy cover by vegetation (%) cover Ground diversity structural Plant height Tree density Tree depth/amount Litter (%) by litter covered Ground debris woody Coarse composition Litter pH) P, N, (C, chemistry Soil penetrability Soil structure Soil content moisture Soil rocks of Presence interactions Inter-specific temperature Ground habitats undisturbed from Distances temperature) (rainfall, Climate Pollution PreUse Habitat Ctry Ants Beetles Spiders arthropods Other traps Pitfall H Authors Clr Shrub, AUSXXXX X 12yrs* 5 - ynny Webb et al. 2000 Grass Clr, Waste RipFor, USAXXXX X X 6yrs 1 - p Williams 1997 Marsh Road RainFor AUS X ?X ?X X X 4-6yrs 2 - ypn n Jansen 1997 Waste Grass ZAF X X 3-9yrs 2 1 y van Hamburg et al. 2004 Weed SclFor AUS X ?X ?X X X 0-4mths 1 - p Pik et al. 2002

Total (incl. '?X') 36 26 19 28 39 16 15 4 2 17 Total (y) 188116542166213111042101 Total (p) 201268764185414333022341 Total (n) 51301005078603615205100 1 PreUse (previous use of the habitat): Mine, minesite, tailing or quarry site; Agr, agriculture (pasture, cropland); Clr, cleared; Weed, weed infested; Waste, waste dump (e.g. ash, dredged material); Road, disused road (pavement removed). 2 Habitat (surrounding habitat): DuneFor, dune forest; Grass, grassland (incl. forbland); Heath, heathland; Marsh, salt marsh; MontFor, montane forest; RainFor, rainforest; RipFor, riparian forest; SclFor, sclerophyllous forest; Shrub, shurubland; TemFor, temperate deciduous/coniferous forest. 3 Ctry (country): AUS, Australia; BRA, Brazil; CDN, Canada; COL, Colombia; DEU, Germany; ESP, Spain; GBR, United Kingdom; ITA, Italy; NLD, Netherlands; NZL, New Zealand; USA, United States of America; ZAF, South Africa. 4 Studies are categorised into chronosequence (investigation of a number of restoration sites representing various age of restoraiton) or longitudinal (long-term study of certain restoration sites) type studies. Year represents: the range of restoration age (years since restoration commenced) for chronosequence studies; or the sampling period for longitudinal studies. * Sampling site(s) represented by 5 Number of true spatial replicates in undisturbed or disturbed reference habitats. 6 'y', factor was explicitly tested by statistical test and/or graphical means and was found to influence colonisation patterns of arthropods; 'p', factor was suggested important based on anecdotal evidences only. 'n', factor was investigated but found non-significant. Only factors that were investigated by more than one study are presented here. ? denotes that the information could not be confidently determined due to insufficient description given in that litereture.

196

APPENDIX 2

Detailed description of the study sites

197 Appendix 2 Detailed description of the study sites. See Figure 2.1 for their locations. Site code B'line Field Size of rainforest History of Grazing Distance from Facing habitat Site name survey exp. Co-ordinates remnants rainforest use Rainforest type Pasture varieties intensity Soil type Altitude boundary§ direction† Note Rob's S1 E1 26°43'1"S, > 10 ha (abutting Selectively Complex notophyll Kikuyu, clover, Paspalum spp. Heavy Basalt and 385-350m P100m, West logged till the (Rhodes grass along the 152°51'1"E sclerophyll forest) 1970's vine forest rainforest edge) colluvium R40m Merg's S12 D1 26°40'4"S, > 10 ha (abutting unknown Complex notophyll Kikuyu, clover, Paspalum Heavy Bassalt 375-400m P>100m, - 152°45'1-4"E sclerophyll forest) vine forest spp. R30m Ray's S2 E2, D2 26°47'1-2"S, 1.15 ha Regrowth since Complex notophyll Kikuyu, clover Heavy Bassalt 515-490m P>100m, NNE

152°48'2"E the 1910's vine forest R22m David's S3 E3 26°50'0-1"S, > 10 ha (abutting Selectively Notophyll feather Kikuyu Light Bassalt and 500-530m P100m, NE Habitat edges of pasture and logged till the metamorphic 152°45'3-4"E sclerophyll forest) 1960's palm vine forest rocks R>30m rainforest on a steep cliff Ed's n/a D3 26°45'0"S, > 10 ha (abutting unknown unknown Kikuyu, clover Heavy Bassalt 425m --Patch of restored forest 152°51'3"E sclerophyll forest) (<0.025) in pasture Phil's S4 E4, D4 26°46'5"-47'0"S, > 10 ha (abutting Selectively Notophyll feather Kikuyu, clover, carpet grass Heavy Bassalt and 525-435m P>100m, NE - NW logged till the 152°46'5"-47'1"E sclerophyll forest) 1960's palm vine forest (sp unknown) colluvium R>30m Richard's S5 E5 26°46'2-3"S, ~11 ha Old regrowth Complex notophyll Kikuyu, clover Heavy Bassalt and 525-455m P100m, North metamorphic 152°47'2-4"E (age unknown) vine forest rocks R>30m Colin's n/a D5 26°46'0"S, as above as above as above Kikuyu, clover Heavy Bassalt 490 m --

152°47'4"E Graham's S6 n/a 26°44'0"S, > 10 ha (abutting unknown Complex notophyll Kikuyu, Rhodes grass?, Heavy Bassalt and 270-250m P>100m, - Small patches of trees (<0.001 ha) in P, rainforest with lantana infestation 152°53'3"E sclerophyll forest) vine forest clover colluvium R30m along edge Neil's S7 n/a 26°46'0-1"S, ~15 ha unknown Complex notophyll Kikuyu, clover Heavy Bassalt 515-480m P>100m, - Pastures with patches of

152°48'4"E vine forest R>30m trees (<0.001 ha) John's S8 n/a 26°47'0-1"S, > 10 ha (abutting unknown Complex notophyll Rhodes grass None Bassalt 395-350m P100m, - Lantana infestation along (occasionary forest edge, no livestock

152°53'0-1"E sclerophyll forest) vine forest mowed) R>30m grazing during the sampling Gary's S9 n/a 26°46'4"S, 1.25 ha Regrowth since Complex notophyll Kikuyu Heavy Bassalt 460m P100m, -

152°49'0-1"E 1902 vine forest R30m Jeff's S10 n/a 26°49'4"S, > 10 ha (abutting unknown Complex notophyll Kikuyu Heavy Bassalt 480-530m P100m, - 152°45'1-2"E sclerophyll forest) vine forest R>30m Paul's S11 n/a 26°47'1"S, > 10 ha (abutting unknown Complex notophyll Kikuyu? Heavy Bassalt 420-480m P>100m, - Rainforest remnant on a

152°48'3-4"E sclerophyll forest) vine forest R>30m steep and rocky slope § Distances from the sampling transects in pasture (P) or rainforest (R) reference sites used for the baseline survey to their closest habitat boundaries. † Directions of rainforest edges facing pasture, where the experimental plots (E1-E5) were established.

APPENDIX 3

Checklist of ‘coarse’ arthropod taxa (Appendix 3a) and ant species (Appendix 3b) found in the reference sites, experimental plots and raw inoculum

199 Appendix 3a Checklist of coarse arthropod taxa found in the reference sites, experimental plots and raw inoculum (see text for more details). Taxa found by pitfall trapping and litter extraction are denoted by □ and ◇ respectively. Taxa of significant rainforest/pasture indicators (Chapter 3) are shown first. Wood = woodchip mulch.

Reference habitats Experimental plots Distance -30 m (within 0 m15 m 100 m 400 m Habitat indicatorsa Shading 90% None (0%) 50% 90% 90% Pitfall Litter Mulch Deep Wood No mulch Shallow Deep Shallow Deep Shallow Deep Shallow Deep Deep Deep Deep Deep Wood a a

Taxa traps extraction RF P (control) (control) Hay Hay Wood Wood Wood Wood Wood Wood Wood Wood Wood (inoculated) Inoculum (litter extraction only) Opilionida RF S RF S □◇ □ ◇ Archaeognatha RF S - □□□□□ Blattodea RF I RF I □◇ □◇ □◇ □ □ ◇ □ ◇ □ □ □ ◇ □ ◇ □ ◇ □ ◇ □ □◇ □ □ ◇ ◇ Coleoptera RF I RF I □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ ◇ Dermaptera RF I RF I □◇ □◇ □◇ □ □ ◇ □ ◇ □ ◇ □ ◇ □ ◇ □ ◇ □ ◇ □ ◇ □ ◇ ◇ □◇ □◇ ◇ Diplopoda RF I RF I □◇ □◇ □◇ □◇ □◇ □ □ □ □ □ □ ◇ □ ◇ □ □ □ ◇ Diplura RF I RF I □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ ◇ Heteroptera RF I RF I □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □ □ ◇ □ □◇ ◇ Isopoda RF I RF I □◇ □◇ □◇ □ □ ◇ □ ◇ □ ◇ □ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ ◇ Pseudoscorpionida RF I RF I □◇ ◇ □◇ □◇□◇ □ □ □◇□◇ □◇ □◇ □ □ □◇ ◇ Psocoptera RF I - □◇ ◇ ◇ □ □ ◇ ◇ □ □ □◇ □ □ ◇ Amphipoda - RF I □◇ □◇ □◇ □ □ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ ◇ Chilopoda - RF I □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ ◇ Formicidae - RF I □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ ◇ Other Hymenoptera - RF I □◇ □◇ □◇ □◇ □◇ □ □ □ □ □◇ □ □ □ □ □ □ ◇ Pauropoda - RF I □◇ ◇ □ ◇ □ ◇ □ ◇ □ □◇ □◇ □◇ □◇ □◇ □◇ ◇ ◇ ◇ ◇ Symphyla - RF I □◇ □◇ □◇ ◇ □ ◇ ◇ ◇ ◇ ◇ Orthoptera P I P I □◇ □◇ □ □ ◇ □ ◇ □ ◇ □ □ ◇ □ □◇ □ □ □ □ □ □ ◇ Araneae P I - □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □ □ ◇ □ ◇ □ ◇ ◇ Homoptera P I - □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □ □ ◇ □ □ □ □ □ ◇ Isoptera - - ◇□ Lepidoptera - - □◇ □◇ □ □ □ □ □ □ □ □ □ □ Onychophola - - ◇□□□ Ostracoda - - □◇ ◇ Protura - - ◇◇ □ ◇ ◇ ◇ ◇ ◇ ◇ ◇ ◇ ◇ Scorpionida - - □◇ □ Siphonaptera - - □□ □ □ Strepsiptera - - □ Thysanoptera - - □◇ □◇ □◇ □◇ □◇ □◇ □◇ □ □ □ □ □ □ □◇ ◇ Diptera (litter extraction only) - - ◇◇ ◇ ◇ ◇ ◇ ◇ ◇ ◇ ◇ ◇ ◇ a RF, P, rainforest and pasture; S, I, habitat 'specialists' and 'increasers' respectively (see Chapter 3 for more details). Appendix 3b Checklist of ant species found in the reference sites, experimental plots and raw inoculum (see text for more details). Species found by pitfall trapping and litter extraction are denoted by □ and ◇ respectively. Species of significant rainforest/pasture indicators (Chapter 3) are shown first. Wood = woodchip mulch. Reference habitats Experimental plots Distance -30 m (within RF) 0 m 15 m 100 m 400 m Habitat indicatorsb Shading 90% None (0%) 50% 90% 90% Pitfall Litter Mulch Deep No mulch Shallow Deep Shallow Deep Shallow Deep Shallow Deep Deep Deep Deep Deep Wood a b b Wood (inoculated Species FG traps extraction RF P (control) Hay Hay Wood Wood Wood Wood Wood Wood Wood Wood Wood (litter Inoculum only) extraction Monomorium tambourinense CCS RF S RF S □◇ □ □ ◇ Pheidole QM2 (ampla grp.) GM RF S RF S □◇ □◇ □ □◇ Anonychomyrma QM3 DD RF S - □◇ □ □ □ □ □ □ □ □ ◇ Leptomyrmex erythrocephalus TCS RF S - □□ □ rufithorax Notostigma foreli SC RF S - □□ Pachycondyla QM2 (porcata grp.) SP RF S - □◇ □ □ Pheidole QM1 (ampla grp.) GM RF S - □◇ □ □ □ □ □ □ □ □ □ Pheidole sp.2 (variabilis grp.) GM RF S - □◇ □ □ □ ◇ Prolasius QM2 (bruneus grp.) CCS RF S - □◇ Carebara QM2 C - RF S □◇ □ □ ◇ Discothyrea QM1 C - RF S ◇ ◇ Hypoponera QM5 C - RF S □◇ □◇ ◇ Lordomyrma QM1 TCS - RF S □◇ ◇ ◇ Mayriella abstinens complex TCS - RF S □◇ ◇ Strumigenys harpyia C - RF S ◇ ◇ Notoncus capitatus CCS RF I - □◇ □ □ □◇ □ □ □ □ □ □ Rhytidoponera chalybaea O RF I - □◇ □ □ □ □ ◇ Hypoponera sp.1 C - RF I □◇ ◇ □◇ □◇ ◇ □ □◇ □◇ □◇ ◇ ◇ ◇ ◇ Paratrechina QM2 (vaga grp.) O P S P S □◇ □ □ □ □ Pheidole QM3 (grp. C) GM P S P S □◇ □◇ □◇ □ □ □◇ □◇ □ □ □ □◇ □ □◇ Cardiocondyla nuda OP S- □◇ □◇ □ ◇ □◇ □ Polyrhachis angusta SC P S - □ Rhytidoponera metallica OP S- □◇ □◇ □ □ □ □ □ □ □ □ □ □ □ □ Carebara QM1 C P I P I ◇ □◇ □◇ □◇ □◇ □◇ □◇ □ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ ◇ ?Anisopheidole sp. C- - ◇□◇ □ □ ◇ □ □◇□◇ □◇◇ Aenictus sp. 1 (ceylonicus grp.) TCS - - □ Amblyopone sp.1 C- - □ Anillomyrma QM1 C- - □◇ □ □◇ □◇ □ □ □ □ □ □ □ Anonychomyrma sp.2 DD - - □ □ Aphaenogaster longiceps O- -□◇ □ □ Reference habitats Experimental plots Distance -30 m (within RF) 0 m15 m 100 m 400 m Habitat indicatorsb Shading 90% None (0%) 50% 90% 90% Pitfall Litter Mulch Deep No mulch Shallow Deep Shallow Deep Shallow Deep Shallow Deep Deep Deep Deep Deep Wood a b b Wood (inoculated Species FG traps extraction RF P (control) Hay Hay Wood Wood Wood Wood Wood Wood Wood Wood Wood (litter Inoculum only) extraction Aphaenogaster pythia O- - □ □ Camponotus sp.1 (rubiginosus grp.) SC - - □ Carebara QM3 C- - □ Cerapachys sp. 1 (turneri grp.) SP - - ◇ Colobostruma biconvexa SP - - ◇ Crematogaster QM1 GM - - □□□ □ Crematogaster QM2 GM - - □ (queenslandica grp.) Crematogaster QM6 GM - - □◇ □□◇□ □ □ □ (queenslandica grp.) Eurhopalothrix australis C- -□◇ □◇ Heteroponera imbellis CCS - - □◇ □◇ □ □ □ □ □◇ □ □ ◇ Hypoponera QM3 C- - □◇ Hypoponera QM4 C- - □◇ ◇◇ Iridomyrmex QM1 (gracilis grp.) DD - - □ □□ □ Iridomyrmex QM3 DD - - □□□□□ □ □ Iridomyrmex rufoniger DD - - □ □ ?septentrionalis Iridomyrmex sp. 1 (mattiroloi grp.) DD - - □ Iridomyrmex sp. 2 (mattiroloi grp.) DD - - □ Leptogenys ?QM1 SP - - ◇ □ Leptogenys hackeri SP - - □□ □□ Leptogenys mjobergi SP - - □□ □□ Leptogenys sjostedti SP - - □□ □ Leptogenys sp.1 SP - - □◇ □ □□□ Leptogenys sp.2 SP - - □ Leptomyrmex nigriventris tibialis TCS - - □ Leptomyrmex QM2 TCS - - □ □□□ □□□ Machomyrma dispar C- - □ Mayriella spinosior TCS - - □□ □◇ □ □ □◇ □◇ □ □ □ Meranoplus QM2 HCS - - □ Meranoplus QM4 HCS - - □◇ □□ Meranoplus sp.1 HCS - - □□◇□ Reference habitats Experimental plots Distance -30 m (within RF) 0 m15 m 100 m 400 m Habitat indicatorsb Shading 90% None (0%) 50% 90% 90% Pitfall Litter Mulch Deep No mulch Shallow Deep Shallow Deep Shallow Deep Shallow Deep Deep Deep Deep Deep Wood a b b Wood (inoculated Species FG traps extraction RF P (control) Hay Hay Wood Wood Wood Wood Wood Wood Wood Wood Wood (litter Inoculum only) extraction Monomorium fieldi formB (nigrium GM - - □ grp.) Monomorium fossulatum C- - □ Monomorium rubriceps TCS - - □ Myopias tasmaniensis C- - □ Myrmecia brevinoda SP - - □ □□ Myrmecia nigrocincta SP - - □□ Myrmecina QM1 TCS - - □◇ □◇ Notoncus ectatommoides CCS - - □◇ □ Ochetellus sp.1 (glaber grp.) O- - ◇ □□□◇ Orectognathus robustus SP - - ◇ ◇ Pachycondyla australis SP - - ◇□ □□ □□ □ □ □ Pachycondyla QM4 (porcata grp.) SP - - □ Paratrechina QM1 (vaga grp.) O- - □ □ □◇ □◇ □ □◇ □ □ □◇ □◇ □◇ □ □ Paratrechina QM3 O- - □◇ □◇ ◇ Paratrechina QM4 O- - ◇◇ Paratrechina QM6 (minutula grp.) O- - □ Pheidole megacephala GM - - □ □ Pheidole QM7 (longiceps grp.) GM - - □ □◇ □ □ □ □ Pheidole QM8 (mjobergi grp.) GM - - □ □□◇□◇□◇□□□□ □ □ Pheidole QM9 (grp. K) GM - - □ □ Pheidole sp.1 (variabilis grp.) GM - - □◇ □ □□◇□□□□□□□ Polyrhachis sidnica SC - - □ Ponera leae C- - ◇ ◇◇ ◇ ◇◇ Pristomyrmex wheeleri TCS - - □ Proceratium ?pumilio C- - ◇ Proceratium sp. C- - ◇ Prolasius QM1 CCS - - ◇ Pyramica membranifera C- - ◇ Rhopalothrix orbis C- - ◇ Rhytidoponera croesus O- - □□ Rhytidoponera victoriae O- -□◇ □◇ □ □◇ □ □◇ □◇ □ □◇ □◇ □◇ □ □ □ □◇ ◇ Solenopsis QM1 C- -□◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ □◇ ◇ Sphinctomyrmex QM1 C- - ◇ Reference habitats Experimental plots Distance -30 m (within RF) 0 m15 m 100 m 400 m Habitat indicatorsb Shading 90% None (0%) 50% 90% 90% Pitfall Litter Mulch Deep No mulch Shallow Deep Shallow Deep Shallow Deep Shallow Deep Deep Deep Deep Deep Wood a b b Wood (inoculated Species FG traps extraction RF P (control) Hay Hay Wood Wood Wood Wood Wood Wood Wood Wood Wood (litter Inoculum only) extraction Sphinctomyrmex sp.1 C- - □ Sphinctomyrmex sp.2 C- - □ Stigmacros QM1 (rufa grp.) CCS - - □◇ Stigmacros QM2 CCS - - □ Strumigenys deuteras C- -□◇ □ □□□◇ Strumigenys nr. cingatrix C- - □ Strumigenys perplexa C- -□◇ ◇ Strumigenys zygon C- - ◇ Tapinoma QM1 (minutum grp.) O- - □◇ □◇ □◇ □◇ □ □ □ □ □ Tapinoma QM3 O- - ◇ Tetramorium ?impressum O- - □□□ Tetramorium bicarinatum O- - □◇ Tetramorium turneri O- - □◇ □ □ □ a Functional groups; C, Cryptic species; CCS, Cold climate specialists; DD, Dominant Dolichoderinae; GM, Generalised Myrmicinae; HCS, Hot climate specialists; O, Opportunists; SC, Specialist predators; SC, Subordinate Camponotini; TCS, Tropical climate specialists. b RF, P, rainforest and pasture; S, I, habitat 'specialists' and 'increasers' respectively (see Chapter 3 for more details).

APPENDIX 4

Supplementary figures

All figures are based on the results of the baseline survey (Chapter 3)

205

□ Pasture, Pitfall ■ Rainforest, Pitfall ◇ Pasture, Litter extraction ◆ Rainforest, Litter extraction

Appendix 4a Sample-based rarefaction curves based on (a) coarse arthropods, (b) ant genera and (c) ant species of pasture and rainforest reference sites, with the expected richness function (MaoTau) of EstimateS software (Colwell 2004). Sample-based rarefaction curves represent expected species density, given N samples for each habitat type. Rarefaction curves indicated that coarse arthropod taxon richness was greater in rainforest than pasture regardless of the sampling methods used. In contrast, rarefaction curves based on ant genera and species showed that species richness derived from litter extraction was lower in pasture than in rainforest, whereas pitfall-trap samples indicated little difference between pasture and rainforest in the total numbers of species or genera.

206

C

■ D n i o Ds i n it ito su t ti turb eti rb p e a p nn m c mo ec oC ■ e C □ □ Pasture, Pitfall ◆◇ ◆ ■□ ◇ ■ Rainforest, Pitfall ■◇ ■ ◇ Pasture, Litter extraction S □ R StressStress ◆ Rainforest, Litter extraction

Appendix 4b Non-numeric triangular ordination of pasture and rainforest sites based on ant functional groups. Ant assemblages are classified in relation to competition, disturbance and stress. Each corner of the triangle represents primary functional group community: C, competitive; S, stress-tolerant; and R, ruderal (see Andersen 2000b for more details). Size of the point represents the number of samples (small, 1-4 sites; medium, 5-8 sites; large, 9-12 sites) that belong to that community type. This ordination showed a tendency for more rainforest sites to have stress-tolerant ‘S’ communities, whereas most pasture sites contained ruderal ‘R’ communities.

207