AUSTRALIA’S BIODIVERSITY – RESPONSES TO FIRE

Plants, birds and invertebrates

A.M. Gill, J.C.Z. Woinarski, A. York

Biodiversity Technical Paper, No. 1 Cover photograph credits Group of 3 small photos, front cover: • Cockatiel. The Cockatiel is one of a group of highly mobile birds which track resource-rich areas. These areas fluctuate across broad landscapes in response to local rainfall or fire events. Large flocks may congregate on recently-burnt areas. /Michael Seyfort © Nature Focus • Fern regeneration post-fire, Clyde Mountain, NSW, 1988. /A. Malcolm Gill • These bull ants (Myrmecia gulosa) are large ants which generally build small mounds and prefer open areas in which to forage for food. They are found on frequently burnt sites. Despite their fierce appearance, they feed mainly on plant products. /Alan York. Small photo, lower right, front cover: • Fuel reduction burning in dry forest. This burn is towards the “hotter” end of the desirable range. /Alan York Large photo on spine: • Forest fire, Kapalga, NT, 1990. /Malcolm Gill Small photo, back cover: • Cycad response after fire near Darwin, NT. /Malcolm Gill

ISBN 0 642 21422 0

Published by the Department of the Environment and Heritage

© Commonwealth of Australia, 1999

Information presented in this document may be copied for personal use or pub- lished for educational purposes, provided that any extracts are acknowledged.

The views expressed in this paper are those of the authors and do not necessarily represent the views of the Department, or of the Commonwealth of Australia.

Biodiversity Convention and Strategy Section Department of the Environment and Heritage GPO Box 636 CANBERRA ACT 2601 General enquiries, telephone 1800 803772

Design: Design One Solutions, Canberra Printing: Goanna Print, Canberra Printed in Australia on recycled Australian paper AUSTRALIA’S BIODIVERSITY – RESPONSES TO FIRE Plants, birds and invertebrates

A. Malcolm Gill CSIRO Division of Plant Industry

J.C.Z. Woinarski Parks and Wildlife Commission of the Northern Territory

Alan York State Forests of New South Wales

Biodiversity Technical Paper, No. 1 Introduction to Biodiversity Technical Paper No. 1

This is the first of the new Biodiversity Technical • Malcolm Gill, CSIRO, Canberra, who Series to be published by the Commonwealth addresses the effects of fire events on plant Department of the Environment and Heritage. biodiversity; The series has been initiated to collate and make • John Woinarski, Parks and Wildlife available information on the conservation and Commission, Northern Territory, who has sustainable use of Australia’s biodiversity to all prepared a review of the literature those involved or interested in biodiversity in concerning the impacts of fire on Australian Australian environments. birds and an annotated bibliography which The papers in this publication have been collates references to fire and Australian written by scientists with expertise in fire and its birds; and effect on Australia’s biodiversity. The papers will • Alan York, then of NSW State Forests, who add considerably to the continuing debate on fire assesses whether frequent hazard reduction in Australia. They will increase access to burning is a sustainable long-term information on how major elements of Australia’s management practice with regard to the biodiversity respond to fire and will be of conservation of terrestrial invertebrate significance to land management agencies, land biodiversity. Alan’s work was made possible by managers and policy makers at all levels. The papers have been prepared for the the foresight of NSW State Forest in initiating Biodiversity Conservation Branch of Environment a long-term project in which data was collected Australia, Department of the Environment and on the same sites for some 20 years. Heritage, Canberra by: The papers were originally prepared for Environment Australia in 1996.

3 Long-term effects of repeated prescribed burning on forest invertebrates: managment implications for the conservation of biodiversity 181 Alan York Centre for Biodiversity and Bioresources, School of Biological Sciences, Macquarie University

Acknowledgements 182

Executive Summary 183

1. Introduction 187

2. Methodology 189 2.1 Study Area 189 2.2 Experimental Design 189 2.3 Measurement of Environmental Parameters 192 2.3.1 Understorey Vegetation Structure 192 2.3.2 The Litter Environment 192 2.3.3 Sticks and Logs 192 2.3.4 Insolation 192 2.3.5 The Soil Environment 192 2.4 Terrestrial Invertebrate Communities 193 2.5 Analytical Procedures 193 2.5.1 Treatment, Exposure and Position Effects 193 2.5.2 Inter-relationships Between Environmental Variables 194 2.5.3 Terrestrial Invertebrate Communities 194

3. Results 199 3.1 Environmental Parameters 199 3.1.1 Understorey Vegetation Structure 199 3.1.2 Litter Biomass 202 3.1.3 Sticks & Logs 203 3.1.4 Insolation 204 3.1.5 The Soil Environment 204 3.2 Inter-relationships Between Environmental Variables 204 3.3 Terrestrial Invertebrate Communities 207 3.3.1 Ordinal Diversity 207 3.3.2 Invertebrate Abundance 207 3.3.3 Invertebrate Species Richness 213 3.3.4 Community Composition 217 3.3.5 Community Structure 234 3.3.6 Biodiversity Indicators 240

7 4. Discussion 243 4.1 Habitat Structure 243 4.2 Terrestrial Invertebrate Communities 245 4.2.1 Invertebrate Abundance 245 4.2.2 Invertebrate Species Richness 246 4.2.3 Community Composition 248 4.2.4 Community Structure 251 4.2.5 Biodiversity Indicators 253

5. Conclusions 257

References 261 Forest Research and Development Division State Forests of NSW

LONG-TERM EFFECTS OF REPEATED PRESCRIBED BURNING ON FOREST INVERTEBRATES: Management Implications for the Conservation of Biodiversity

Consultancy Report to the Department of the Environment and Heritage

Alan York

Final Report April 1996 ACKNOWLEDGEMENTS

Full credit must be given to Adrian van Loon Urquart who assisted in the laboratory work and who, in the 1950s, conceived and established a with reference material, and to Traecey Brassil field trial which was sufficiently well designed and and Alison Towerton for their assistance with the robust to not only be addressing questions still data analyses and presentation. Taxonomic relevant 40 years later, but also sufficiently flexible support was provided by specialists Mike Gray, to now facilitate questions probably not even Dan Bickel, John Lawrence, Robert Taylor and conceived at that time. Similarly, much is owed to Gerry Cassis who also provided enthusiastic the Forestry Commission of New South Wales guidance and advice. Thanks also to Andy Beattie (State Forests) and its staff for the long-term and Ian Oliver from the Key Centre for maintenance of the trial, in particular Bill Buckler Biodiversity & Bioresources at Macquarie who methodically and reliably undertook routine University for their ongoing support and counsel, measurements for over 20 years, and to Hugh particularly in regard to the development and use Dowden, Bob Bridges and others who maintained of methodologies for Rapid Biodiversity and verified the enormous database. Assessment. For my part in this project, which began in I also wish to sincerely thank Neal Hardy for 1991, I wish to thank Bill Buckler, Bill Chapman, his patience and perseverance, and the Patrick Murphy and Traecey Brassil for their Department of the Environment and Heritage for assistance with fieldwork, and Rita Holland, Ruth their financial support. Burton, Graeme Price and Darren Waterson for the tremendous effort and dedication involved in sorting in excess of 55,000 specimens. My appreciation also to Debbie Kent and Chris-Ann

182 EXECUTIVE SUMMARY

BACKGROUND THIS REPORT

Infrequent, periodic forest fires (bushfires) are an Little is known about the effects of repeated integral part of the modern physical environment of hazard-reduction burning over long time scales. Australian sclerophyll forests. Low-intensity fires The research reported here was conducted as part are extensively used in managed sclerophyll forests of the Bulls Ground Frequent Burning Study to stimulate regeneration, manipulate wildlife (Experiment F8/2.9), located on the mid-north habitat and in particular, to reduce fuel levels with coast of New South Wales and established in the intention of minimising the extent and severity 1969. The terrestrial invertebrate component of of wildfires. In Australia, the use of deliberate fire to the project commenced in 1991 following 20 years prevent high-intensity wildfires has become of repeated low-intensity fire, and was undertaken probably the most extensive use of fire in land with the following aims: management. The inherent variability in natural fire • to identify the effects of long-term repeated regimes generally results in a mosaic of habitats burning on terrestrial invertebrate with vegetation at different stages of floristic and biodiversity, structural post-fire succession, each potentially • to identify species and/or species groups supporting particular animal communities. Changes most affected by this management practice to the components of the fire regime (fire intensity, and to devise strategies to ensure their frequency and season of occurrence), as a continued conservation, consequence of forest management practices, have • to identify species which, due to the nature the potential to alter the composition and structure of their response, may be useful “indicators” of natural communities. The research reported here of environmental disturbance and deals with the impact of frequent low-intensity fire degradation. (“hazard-” or “fuel-reduction burning”) on the The primary outcome was therefore to assess abundance, richness, composition and structure of whether frequent hazard reduction burning is a terrestrial invertebrate communities. sustainable long-term management practice with regard to the conservation of our forest WHY INVERTEBRATES? biodiversity. Invertebrates (insects, spiders, mites, worms, snails, FREQUENT FIRE & centipedes etc.) are the most diverse and abundant HABITAT STRUCTURE animals in most natural systems, but their importance in sustaining those systems is If frequent fire reduces the diversity of post-fire commonly not appreciated. This multitude of environments, then it has the potential to impact organisms constitutes the bulk of the biodiversity upon animal communities dependent upon this within forests and plays an essential role in primary habitat mosaic. This research indicated that production, nutrient cycling and uptake, population frequent burning resulted in a simplification of & community level interactions and energy storage large-scale spatial patterning in the litter (fine-fuel) & transfer. Through their contribution to environment. The components (leaves, twigs, bark ecosystem function, these organisms also enable etc) that give the leaf litter its physical structure forest ecosystems to provide benefits to humanity. changed with regard to their relative abundance These benefits include amenity values in the form and spatial distribution. There were marked of aesthetics, recreation and education; heritage changes to the amount (cover) of vegetation in the values as forests contribute to long-term security understorey and its spatial patterning. While the for catchment protection, air and water quality and quantity of vegetation closest to the ground nature conservation; and economic values including (ground herbs & small shrubs) was not affected by timber production, grazing and ecotourism. The frequent burning, there was a decrease in the maintenance of biodiversity is a fundamental spatial heterogeneity (patchiness) of these layers. principle underlying the ecologically sustainable Conversely, the cover of tall and very tall shrubs management (ESM) of these environments.

183 Australia’s Biodiveristy - Responses to Fire

was substantially reduced and showed an increase with broad tolerances, or adaptations, to drier and in spatial heterogeneity. Top-soil moisture levels more open environments. were, on average, 18% lower following 20 years of These shifts in community composition were frequent burning, whereas the amount of light substantial and suggested that the extensive and reaching ground level had increased (on average) frequent application of fuel-reduction burning by 125% and become more spatially homogeneous could result in a reduction in terrestrial (less patchy). A number of habitat components (eg. invertebrate biodiversity at a regional scale, with top-soil hardness, the distribution of large sticks & this decrease potentially as high as 50%. Current logs) however showed no significant response to fuel management strategies which limit the extent frequent burning. of frequent burning will ameliorate these impacts, however there remains a need to establish secure TERRESTRIAL INVERTEBRATE refuges for species with specialist requirements and COMMUNITIES limited dispersal abilities, and provide links (ie corridors) between habitat patches to facilitate This study revealed a rich terrestrial invertebrate recolonisation. The effectiveness of similar fauna with representatives from the Chelicerata strategies developed to conserve vascular plants (spiders, ticks & mites, pseudoscorpions, and vertebrates remains untested however for the harvestmen), Crustacea (landhoppers, slaters), groups which actually constitute the bulk of our Chilopoda (centipedes), Diplopoda (millipedes), forest biodiversity. Realistically, the conservation and a diverse array of Insect Orders and Families. of biodiversity cannot be achieved without Numerically, the most abundant groups overall consideration of the important role that were the springtails (33%), ticks & mites (24%) invertebrates play, both through their involvement and ants (23%), with these three groups in ecological processes, and as a significant representing 80% of all individuals caught. For 10 component of the overall richness of biotic broad taxonomic groups there were sufficient data communities. to statistically test the effects of frequent burning. The results indicated a variety of responses with Community Structure & statistically significant decreases in abundance for Ecosystem Function ↓ ↓ ticks & mites ( 31%), insect larvae ( 35%), flies The biological structure of a community involves ↓ ↓ ( 58%) and beetles ( 31%). Many of these groups species composition (diversity and relative are associated with leaf litter and it is likely that abundance) and the relationships between species - their numbers have been influenced by the their ecological role. It was demonstrated here that episodic removal of this resource. Three groups considerable additional detail concerning, and showed substantial increases in abundance insight into, the nature of invertebrate community ↑ following frequent burning; bugs ( 77%), ants changes could be provided by the inclusion of fairly ↑ ↑ ( 250%) and spiders ( 33%), probably as a general information concerning habitat and dietary response to both changes in habitat suitability and preferences. It was apparent that frequent burning increased ease of capture in a simplified leads to a change in the structure of the environment. invertebrate community. Within species Biodiversity assemblages there were shifts based on feeding strategy and habitat preference. While the impact Using ants, beetles, flies, spiders & bugs as of these changes on ecosystem function was beyond representative groups and potential indicators of the scope of this study, substantial measured environmental degradation, this research changes in the structure of invertebrate demonstrated that although overall species assemblages and the loss of species associated with α richness at specific sites ( -diversity) did not the decomposer cycle implies frequent burning change with frequent burning, all groups showed may be impacting upon nutrient cycling and substantial changes in the composition of species transfer within these forests. If this is the case, it assemblages. There was a loss of taxa dependent would have serious implications with regard to the upon a substantial litter layer and stable moist maintenance of ecological sustainability. conditions, and these species were frequently habitat or dietary specialists and often uncommon or “rare”. The overall diversity of frequently burnt areas was maintained by the addition of species

184 Bushfire and forest invertebrates

Biodiversity Indicators Note: Indices used to gauge the success of ecologically Following the preparation of this report, there sustainable management practices need to be have been some taxonomic revisions and interpretable, significant and cost efficient. They associated morphospecies corrections of the ant also need to account for variability in space and data. These have been independently published, time, and be appropriate for the scale of however they were of a minor nature and do not management. The research reported here identified alter the outcomes of the analyses or the the limited usefulness of data obtained using coarse- conculusion drawn in this report. scale taxonomic classification (eg. Family or Order), with the cost-effectiveness of abundance data alone shown to be low. This research also identified substantial limitations with regard to the use of a single index, species richness, as a measure of change and/or environmental impact. Species richness (α-diversity) is frequently used to describe and compare communities, however in this case it was found to provide a deceptive summary of community characteristics and severely restrict the level of interpretation that could be derived for impact assessment purposes. The application of Rapid Biodiversity Assessment (RBA) methodology here demonstrated that the study of the composition and structure of communities is likely to prove more rewarding in this regard. The identification of individuals to distinct “morphospecies” facilitated the incorporation of broad-level ecological information into the assessment, and interpretation, of environmental impact. This in turn enabled the development of management recommendations consistent with the conservation of biological diversity.

185

Bushfire and forest invertebrates

1. INTRODUCTION

The concept of Ecologically Sustainable there is growing concern that repeated low- Development (ESD) was defined by the United intensity burning, as a management prescription, Nations in 1987 as “… development that meets the may have a negative influence on plant and animal needs of the present without compromising the ability of communities. Frequent firing may remove future generations to meet their own needs” vegetation species that rely on seed production for (“Bruntland Report” - WCED 1987). This their persistence (Gill 1981; Bradstock and concept has been developed and refined regularly Myerscough 1981; Benson 1985; Fox and Fox since that time, most recently at the “Earth 1986), often leading to dominance by herbaceous Summit”, the United Nations Conference on fire-tolerant species (Cary and Morrison 1995). Environment and Development held in Rio de Fire frequency becomes a significant factor for Janeiro in 1992. ESD forms part of the World plant species requiring a long period of time Conservation Strategy (IUCN 1980) and is the (relative to the interval between fires) to reach basis for the National Conservation Strategy for reproductive maturity (Zedler et al. 1983; Australia (Commonwealth of Australia 1983). Nieuwenhuis 1987). Changes in habitat structure The conservation of biological diversity is a as a consequence of frequent burning are likely to foundation of ESD and is one of the three core disadvantage many native mammal and bird objectives of the Australian National Strategy for species (Catling 1991; Whelan 1995). Ecologically Sustainable Development. Biological While sclerophyll forests, woodlands and diversity refers to the variety of all life forms - the heaths are dominated by plant species with different plants, animals and micro-organisms, the adaptive responses to fire that enable them to genes they contain and the ecosystems of which survive exposure to periodic burning (see for they form part. Australia has ratified the example Gill 1981; Noble and Slatyer 1981), the Convention on Biological Diversity arising from impact of such fires on terrestrial invertebrates is the Earth Summit, and is now developing strategies poorly understood. The consumption of some or to assess and protect its biodiversity. The all of the leaf litter by flame, short-lived but conservation of biological diversity is a major substantial rises in soil temperature during fire, objective of the National Forest Policy Statement and post-fire changes in the surface radiation (NFPS 1992), to be achieved through the budget, mean that soil and litter fauna are protection of ecosystems (reserve strategies) and substantially affected by fire in the short-term complementary off-reserve management (Bornemissza 1969; Springett 1979; Moulton (Ecologically Sustainable Management — ESM). 1982; Coy 1996). Recovery from a single fire may In New South Wales, State Forests has put forward take up to 3-5 years (Metz and Farrier 1973; ESM as a major objective in its 1992–5 Corporate Seastedt 1984; Neumann and Tolhurst 1991), Plan (Forestry Commission of NSW 1992). This however the timing and intensity of burning is concept has been widely adopted by other land important, as is the mobility and recolonising management agencies throughout Australia and ability of particular species (Morris 1975). Given forms part of the National Strategy for the the patchy nature of low-intensity fuel-reduction Conservation of Australia’s Biological Diversity. burns, and the protection afforded by small Low-intensity fires are extensively used in habitat refuges and within the soil, it has been managed sclerophyll forests to stimulate suggested that periodic fires used for fuel regeneration, manipulate wildlife habitat and in management purposes have few long-term effects particular, to reduce fuel levels with the intention on most soil and litter invertebrates (Majer 1980; of minimising the extent and severity of wildfires. Campbell and Tanton 1981; Abbott et al. 1984). In Australia, the use of deliberate fire to prevent There is little information on the effects of high-intensity wildfires has become probably the fire frequency on forest invertebrates, but Abbott most extensive use of fire in land management et al. (1984) suggest that periodic low intensity (Whelan 1995). While infrequent, periodic fires fires have few permanent effects on most of the (bushfires) are an integral part of the modern invertebrate taxa present in the litter and soil of environment of Australian sclerophyll forests, the Jarrah forest. Long-term studies of spiders

187 Australia’s Biodiveristy - Responses to Fire

(Huhta 1971; Merrett 1976) and ants (York 1994, thereby making a substantial contribution to our 1996) suggest that, in the years following fire, National biodiversity (New 1984; CONCOM there is a replacement series of groups of species 1989). Realistically, the conservation of related to their particular habitat requirements biodiversity cannot be achieved without being met as the habitat changes in structure over consideration of the important role that time. A number of species persist throughout this invertebrates play, both through their involvement period, but show changes in relative dominance in ecological processes, and as a significant within the community. York (1996) suggested that, component of the overall richness of biotic for ants, the use of regular widespread fires for communities. fuel reduction was likely to result in a truncation The research reported here was therefore of these successional patterns and an associated undertaken with the following aims: loss of regional biodiversity. • to identify the effects of long-term repeated Periodic low-intensity fire (hazard-reduction burning on terrestrial invertebrate burning) is a conspicuous management strategy in biodiversity, virtually all of Australia’s dry forest communities. • to identify species and/or species groups While it is primarily used to reduce fuel levels, most affected by this management practice little is known about the effects of its repeated use and to devise strategies to ensure their on natural ecosystems over long time-scales. On continued conservation, the east coast of NSW, extensive wildfires in • to identify species which, due to the nature January 1994 have led to calls for increased use of of their response, may be useful “indicators” hazard-reduction burning, however the impacts of of environmental disturbance and the resulting increase in fire frequency are poorly degradation. understood in the very forest environments this The primary outcome is therefore to assess management strategy seeks to protect. The whether frequent hazard reduction burning is a paucity of information available on the effects of sustainable long-term management practice with increased fire frequency on forest invertebrates is regard to the conservation of our forest of considerable concern. Invertebrates constitute biodiversity. 95% of known species of fauna in Australia,

188 Bushfire and forest invertebrates

2. METHODOLOGY

2.1 STUDY AREA whenever fuel build-up permitted, generally every 3 years (1970, 1973, 1977, 1980, 1983, 1986, As part of the F8 series of “fire effects” studies, 1989, 1992). This burning regime is ongoing. State Forests has an ongoing experimental project A program was instituted to monitor aspects which is particularly suitable for addressing of the response of this forest to repeated low- questions relating to repeated disturbance and intensity fire. A number of parameters were biodiversity conservation. The F8/2.9 Frequent regularly measured on each research plot: tree Burning Study is located in even-aged coastal growth, major and minor understorey vegetation, blackbutt Eucalyptus pilularis regeneration in litterfall, and fine and heavy fuel. These Compartment 70, Bulls Ground State Forest, measurements were made systematically between Kendall Management Area on the mid-north 1970 and 1987, when the project was reviewed; coast of New South Wales (31°33'S, 152°38'E, and then less frequently until 1992. 240m ASL.). The stand was logged and silviculturally treated in 1958–9, with seed trees 2.2 EXPERIMENTAL DESIGN retained singly and in groups, and unmerchantable trees culled in line with Timber From an inspection (by the author) of the area in Stand Improvement (T.S.I.) techniques. The area 1990 it was apparent that twenty years of repeated has experienced no further management burning had resulted in substantial changes in treatment (except experimental fuel-reduction macro- and micro-habitat parameters. It was burning) since that time. hypothesised that these changes would have had a In 1969 twenty-one 0.225 acre (0.1 ha.) significant effect on terrestrial invertebrate temporary plots were established in openings communities. In 1991, two years after the last fire, created by the logging treatment which carried a a project was initiated to assess the impact of good stocking of young blackbutt regrowth long-term fuel reduction burning on terrestrial (11 years old). These areas were found to support invertebrates, and to investigate the possibility of a mean number of 339 stems per hectare (Van using this faunal group as monitoring agents in Loon 1970), consisting mainly of blackbutt (48%) the assessment of ecologically sustainable and bloodwood E. gummifera (31%). The management. The overall approach was to view remainder (21%) consisted most commonly of this single sample period as a “snapshot” of the turpentine Syncarpia glomolifera, red mahogany effects of 20 years of prescribed burning by E. resinifera, white mahogany E. acmeniodes and comparing burnt and unburnt replicates. While grey gum E. punctata. Following an assessment of this does not enable a description of changes over stand parameters, a number of these plots were time, it does provide a unique opportunity to selected on the basis of their similarity for a long- assess the long-term impact of this management term fire study (F8/2.9), which was formally practice. initiated in March 1970. Twelve of the fourteen plots were selected as Fourteen 0.1 ha. permanent research plots suitable, six within each treatment (unburnt & were established, 7 randomly allocated as burning burnt). Plots 7A and 7B were excluded as they treatments (burnt), the remaining 7 as control contained rocky outcrops and were subjectively (unburnt) plots from which fire was excluded assessed to be different to other plots. (7×2 randomised block design). These study plots Randomised assignment of treatments to were located within similarly treated forest blocks experimental units ensured “true” replication of of approximately 1 ha. and separated by cleared treatment effects (see Hurlbert 1984). In order to buffer areas to protect them from wildfire (see increase the sensitivity of the experiment by Figure 2.1). For the remainder of this report the increasing the “precision” with which properties term “plot” refers to the 1 ha. treated forest areas, of each experimental unit (plot) and hence each while “research plot” refers to the 0.1ha study treatment were estimated, it was necessary to take plots defined in 1970 (see Figure 2.2). Fuel multiple samples from each plot. Four 20m reduction burning was implemented in Autumn transects were therefore established within each

189 Australia’s Biodiveristy - Responses to Fire

Figure 2.1 F8/2.Frequent Burning Study, Bulls Ground State Forest. Location of study plots. (Plots 7A and 7B not shown).

190 Bushfire and forest invertebrates

Figure 2.2 Schematic layout of study “plot” plot (a “nested” design), each on a randomly- Table 2.1 Slope and aspect of study plots oriented compass bearing starting from each corner Unburnt Burnt × of the established “research plot”. A 20m 10m sub- Plot Sub-plot Slope° Aspect° Slope° Aspect° plot was then centred on this transect (see Figure 1 1 1 290 1 220 2.2) in order to assess the small-scale variability of 2 1 300 1 220 measured parameters. 3 0 325 2 245 The general physical characteristics of each 4 1 330 3 210 sub-plot were summarised by measurements of 2 1 1 250 2 320 ground slope and aspect. The average slope of the 2 3 240 5 300 site in degrees below the horizontal was 3 3 260 2 340 determined with a hand-held clinometer, while the 4 0 360 3 310 aspect was determined by use of a compass. The 3 1 2 290 2 310 sub-plots had low slope angles (0–9) and 2 4 330 2 315 predominantly north-west to south-west aspects 3 1 290 1 290 (see Table 2.1). On average, burnt sites had slightly 4 0 280 4 310 steeper slopes, primarily plots 4, 5 & 6. The 4 1 1 210 2 250 differences in slope however were slight and reflect 2 3 240 9 285 the ridge-top nature of the study area. The range of 3 3 195 5 265 aspects was similar for both treatments. 4 1 200 5 270 5 1 1 225 2 275 2 2 280 4 270 3 0 270 2 255 4 2 275 6 270 6 1 0 360 7 225 2 1 230 7 200 3 1 210 2 200 4 1 220 3 220 Range 0–4 195–360 1–9 200–340 Mean±s.e. 1.4±0.2 269±10 3.4±0.4 26±69

191 Australia’s Biodiveristy - Responses to Fire

2.3 MEASUREMENT OF Table 2.2 Structual vegetation components ENVIRONMENTAL PARAMETERS Height class Structural component 0–20 cm Ground herbs The environmental framework within which 20–50 cm Small shrubs terrestrial invertebrate communities function 50–100 cm Mid-sized shrubs primarily involves elements of the vegetation 100–150 cm Tall shrubs understorey, the top-soil and litter components. A 150–200 cm Very tall shrubs number of parameters were quantitatively assessed to evaluate their possible influence on species richness and community structure. 2.3.2 The Litter Environment The distribution of data for most variables suggested that the sample mean was the best Ground-dwelling invertebrates have been shown estimate of average conditions at each sub-plot. to be sensitive to levels of forest litter, particularly Because individual samples were randomly drawn during post-fire recovery (Bornemissza 1969; from within replicates, a measure of variability Springett 1976; Seastedt 1984). Five randomly 2 about the mean also provided information about placed samples (0.1 m ) of litter (including sticks the spatial variability (“patchiness”) of the variables up to 2.5cm diameter) were collected from each concerned. The coefficient of variation (CV = sub-plot, sieved with 1mm soil sieves to dislodge standard deviation/mean × 100%) was selected as soil material, and then dried in an oven at 105° for the most appropriate measure here due to the 72 hours. Material was then sorted into 5 large fluctuations in mean values and the observed components and weighed: twigs 0–5mm & twigs dependence of the standard deviation on the mean. 6–25mm diameter, bark, leaves, and very fine fuel In order to satisfy the assumptions underlying (miscellaneous decomposing matter). This particular statistical procedures, variables were approach was consistent with that used to estimate appropriately transformed as required. the “fine fuel” fraction over the previous 20 years. 2.3.1 Understorey Vegetation Structure 2.3.3 Sticks and Logs The physical structure of the vegetation The incidence and diameter of all sticks & logs understorey in this forest environment consists (“heavy fuel”, >2.5cm) was recorded along 2 primarily of a shrub stratum and a herb stratum orthogonal 20m transects centred on each sub- whose heights and spatial distribution are a function plot. Following an examination of the frequency of fire history. A structural classification of the distribution of values, data were grouped for vegetation was chosen because it allowed a relatively subsequent analyses into the following five quick and consistent assessment of the sites (48 sub- diameter categories: 2.5–9.9, 10–24.9, 25–49.9, plots in total) to be made in an environment which 50–74.9, 75+cm. is floristically diverse (Doug Binns pers. comm.). 2.3.4 Insolation Vegetation structure is of direct significance in ecological studies of soil and soil-surface Levels of insolation have been shown to be critical invertebrates because the amount and distribution factors determining the abundance and of vegetation determines both the physical distribution of certain terrestrial invertebrates. framework within which activity takes place, and the The amount of light reaching the forest floor was food availability and hence carrying capacity of the used as an index of insolation levels, and measured environment (Greenslade and Thompson 1981). using a Lunasix 3 Gossen exposure meter, fitted Vegetation structure was quantitatively with incident light cone. Twenty measurements assessed using the “cover-board” technique (see were taken systematically within each sub-plot MacArthur & MacArthur 1961; Fox 1979). and expressed as a percentage of available light as Percentage cover was measured at 20 points measured outside the forest at that time of day. systematically located along each transect for five 2.3.5 The Soil Environment structural components of the understorey (see Table 2.2). The mean of the 20 measurements was The underlying geology of the site consists used as an estimate of percentage cover for each primarily of conglomerate, sandstones and shales. vegetation layer at each sub-plot, and the These have weathered to form shallow soils coefficient of variation (CV) as an estimate of (yellow earths & brown podzolics) which are spatial variability. relatively low in nutrients. Two aspects of the soil

192 Bushfire and forest invertebrates

physical environment were assessed: top-soil beetles (Coleoptera), spiders (Araneae), bugs moisture and top-soil hardness. (Hemiptera) and flies (Diptera) were subsequently 2.3.5.1 Top-soil Moisture sorted to “morphospecies” using the protocols Five samples (10cm diameter 3cm deep) per sub- described in Oliver & Beattie (1993), with final plot were collected and kept in sealed containers. taxonomic verifications being performed by Mike Samples for each sub-plot were pooled in the Gray (spiders), Gerry Cassis (bugs) and Dan laboratory, weighed and dried in an oven at 105° Bickel (flies) of the Australian Museum, John for 72 hours. An estimate of “field moisture Lawrence (beetles) and Robert Taylor (ants) of the content” for each sub-plot was calculated in the CSIRO. following manner (see Lambert 1982): Oliver and Beattie (1996a) have shown that morphospecies can provide a robust estimate of % Moisture Content = species richness across a variety of habitats. This study provides a substantial test of the hypothesis air-dried weight - oven dried weight × 100 air-dried weight that the lack of knowledge concerning so many Australian invertebrates, the so called “taxonomic impediment” (Taylor 1983), no longer prevents 2.3.5.2 Top-soil Hardness the inclusion of invertebrates in biodiversity An index of top-soil hardness (0–5cm) was assessment and studies of management impacts. obtained using a Geonor inspection vane, which measures soil shear strength. Twenty 2.5 ANALYTICAL PROCEDURES measurements were taken within each sub-plot, the mean value representing the average shear All initial analytical procedures were performed strength and the coefficient of variation (CV) an using the SPSS statistical package on a 486PC at indication of spatial variability. SFNSW’s Research Division. Data distributions were examined using exploratory data analysis 2.4 TERRESTRIAL INVERTEBRATE techniques (EXAMINE) and transformed (as COMMUNITIES required) for subsequent analyses (MANOVA and REGRESSION). Canonical Correspondence Epigaeic (surface active) invertebrate communities Analyses (CCA - Ter Braak 1986) were performed were assessed by a single summer pitfall trapping using programs written in Splus on a Sun program in February 1991. Nine points were Workstation. established and marked along the 20m transect within each sub-plot. At each point a 6.5cm 2.5.1 Treatment, Exposure diameter 9cm deep plastic cup was sunk flush with and Position Effects the ground surface and half-filled with a non- Plots had been allocated to one of two treatments: attractive preservative solution. Pitfall traps were burnt (1B-6B) or unburnt (1A-6A). Aspect values left open for a period of 7 days (5–12th February were coded from 1-6 to reflect the relative 1991), reducing the effect of temporal changes in “exposure” of sub-plots to solar radiation; with abundance and activity on estimates of species 300–330° = 1 (highest), 270–300° = 2, richness and community composition (York 1989). 330–360° = 3, 240–270° = 4, 210–240° = 5, and Weather during this period was typical for that 180–210° = 6 (lowest). Sites intermediate between time of year; temperatures ranged from 17–35°C categories were allocated an average (mean) code. and 27mm of rain fell between the 7th and 8th. To evaluate any large-scale spatial trend in Samples were returned to the laboratory and habitat variables (and species’ responses) a new examined with a binocular microscope where variable (position) was generated to reflected the material was sorted to the taxonomic level of north-south location of plots along Sandy Order. A number of groups were chosen for more Hollow Road (see Figure 2.1). The value of aspect detailed investigation based on the criteria of and position for each plot ranged from 1-6 (see sufficient numbers for statistical analysis, the Table 2.3). ability to recognise and define species, and the likely appropriateness of the sampling methodology. Ants (Hymenoptera: Formicidae),

193 Australia’s Biodiveristy - Responses to Fire

Table 2.3 Values of exposure and position for component delineates the largest pattern of each sub-plot. Aspect was coded from 1-6 to relationships in the data (defines the greatest reflect the relative exposure of sub-plots to solar amount of variation in the data); the second radiation (see text). Position was coded from 1 delineates the next largest pattern and so on. (north) to 6 (south) to reflect location along Sandy 2.5.3 Terrestrial Invertebrate Hollow Road (see Figure 2.1) so as to detect Communities possible spatial patterns. Samples were sorted to Order using Exposure (sub-plots) morphological characteristics and general Treatment Plot Position 1 2 3 4 taxonomic keys. Relative abundance of individuals 1A 1 2 1.5 1 2 within these groups at plots differing in treatment 2A 2 4 4.5 4 3 and position were examined using Analysis of Unburnt 3A 3 2 2 2 2 Variance (ANOVA) procedures. 4A 5 5.5 4.5 6 6 5A 4 5 2 3 2 2.5.3.1 Biodiversity Selected taxa (see 2.4) were described in terms of 6A 6 3 5 5.5 5 the relative abundance of individuals within 1B 2 5 5 4 5.5 constituent groups (families, sub-families, genera 2B 1 1 1.5 3 1 etc. as appropriate), and their species richness (as Burnt 3B 3 1 1 2 1 defined by morphospecies). Patterns in these 4B 4 4 2 4 3 community descriptors at plots differing in 5B 5 2 3 4 3 treatment and position were examined graphically, 6B 6 5 6 6 5 and using Analysis of Variance (ANOVA) procedures. Patterns in environmental parameters at plots differing in treatment and position were 2.5.3.2 Community Composition examined graphically, and using Analysis of Patterns of species’ responses to treatments are Variance (ANOVA) procedures. For frequency illustrated in tables of relative abundance. This data, the degree of association between variables enabled broad “assemblages” of species, with was examined using contingency tables similar responses to disturbance, to be identified. (crosstabulation), with significant associations 2.5.3.3 Environmental Determinants of tested using the χ2 statistic. Community Composition 2.5.2 Inter-relationships Between The relative importance or ability of the measured Environmental Variables habitat variables to explain the composition of invertebrate assemblages was assessed using The environmental (habitat) variables were Canonical Correspondence Analysis (CCA, Ter subsequently analysed using an ordination Braak 1986, 1991). This method arranges species procedure (Principal Components Analysis - along environmental gradients by constructing PCA) in order to untangle linear relationships linear combinations of environmental factors between variables, and reflect inherent structural which result in maximal separation of species’ patterns. In this analysis, each pattern appears as a distributions in ordination species-space. These component delineating a distinct cluster of analyses were performed to determine whether interrelated data. Components are rotated any differences in turnover or spatial similarity of orthogonally (VARIMAX procedure) to clarify the assemblages among taxa might be explained by the definition of these clusters by maximising or different taxa responding to environmental minimising correlations between variables and gradients. Ter Braak (1986) fully describes the components. The projection (the loadings) of underlying assumptions and strengths of this each variable on the component axes defines the method. The main assumption is that individual clusters of variables. Kaiser’s criterion (only the species response models are all similar and of components with eigenvalues greater than one) unimodal, Gaussian form. Although it is doubtful was used to determine the number of components whether this assumption is reasonable for all to be extracted. Eigenvalues measure the amount species, CCA has been shown to be robust to of variation accounted for by a pattern, while the moderate violations of assumptions (Palmer 1993) loadings measure the degree to which variables and offers the potentially most powerful method are involved in the component pattern. The first

194 Bushfire and forest invertebrates

available in revealing patterns of community Broad groups of this kind (functional groups) were composition in relation to environmental factors. identified in this analysis by reference to the It also has the advantage that the results are literature (eg. see Andersen 1990) and following unaffected by correlations among environmental discussions with relevant taxonomists. The variables. numbers of morphospecies within these groups Results of the CCA ordination were was graphically presented and examined in order displayed as “bi-plots” which show the to detect those which may be sensitive to configuration of the variables, the scatter of sub- microhabitat features associated with structural plots, and the relationship between the two. This characteristics of the environment. gives an overview of how community composition Additionally, the relative abundance of varies with the environment (Ter Braak 1986). species recorded in one treatment only was The interpretation of the results of the bi-plots displayed in tabular form and discussed in relation was simplified by using a sub-set of the to their likely ecological roles. environmental variables in the analyses. This sub- 2.5.3.5 Biodiversity Indicators set was composed of representative variables from Decisions regarding conservation evaluation often each of the eight independent patterns identified are based upon the diversity (species richness) of by the Principle Components Analysis (see 2.5.2), the area under concern (see Margules et al. 1988). with the additional inclusion of two largely Similarly, species richness is a common independent variables: aspect and insolation. “performance indicator” used for monitoring 2.5.3.4 Community Structure disturbance impacts (see Kremen 1992). In order Analyses of species’ “assemblages” often fails to to simplify these processes, it is often adequately account for rare species, which are hypothesised that one taxonomic or functional frequently represented by too few records to allow group may reflect the response of other taxa, and any meaningful patterns to be determined (see hence function as “indicator” or “umbrella” York 1994). One common means to overcoming species. this problem, at least to some extent, is to group To test whether the richness of particular species according to some ecological invertebrate taxa could be useful in predicting characteristic, so that the collective behaviour of overall invertebrate biodiversity, the relationship the group can be assessed. At the species level between species richness of selected taxa was there is insufficient ecological information for investigated using correlation analyses (Pearson’s most groups to do this with confidence, however Product-Moment and Spearman’s Rank). broad grouping may be identified at higher taxonomic levels, such as sub-family or family.

195 Australia’s Biodiveristy - Responses to Fire

Forest area one day after a low-intensity fuel-reduction burn. The small areas of leaf litter and unburnt understorey vegetation remaining indicate the patchy nature of such burns. These represent potential refuges for terrestrial invertebrates; refuges that are reduced in number and extent by frequent fire. /Alan York

Dry eucalypt forest that has remained unburnt for over 25 years. These forests are characterised by deep leaf litter and low light levels. /Alan York

196 Bushfire and forest invertebrates

Dry eucalypt forest that has been frequently burnt for the past 25 years. These forests are very open and characterised by low leaf litter levels and high light levels. /Alan York

A spider from the Family Salticidae. These “jumping spiders” hunt for their food on understorey vegetation, trees and logs. The two species from this Family were only found on unburnt plots, suggesting that they prefer habitats with more structurally complex vegetation in A spider from the Family Zodariidae. These spiders which to hunt. /Alan York typically live under stones, logs and in leaf litter. There were four times as many species from this Family on frequently burnt plots, suggesting that they prefer these more open habitats in which to hunt. /Alan York

197 Australia’s Biodiveristy - Responses to Fire

An ant from the genus Probolomyrmex. It is a rare “cryptic” species that was only found in leaf litter samples collected from one site. Little is known about its habitat preferences, although it is not thought to be disadvantaged by frequent burning at this stage. /Alan York

A spider from the Family Lycosidae. These “wolf spiders” are ground hunters. All three species from this Family were only found on frequently burnt plots, suggesting that they prefer these more open habitats in which to hunt. /Alan York

This is a species of ant known as Rhytidoponera An ant from the genus Orectognathus. These ants are metallica which is an “opportunist” commonly found in specialist predators who use their long mandibles to disturbed habitats. In this study it was 500 time more catch soft-bodied insects such as Springtails abundant on frequently burnt sites, potentially indicating (Collembola). Because of their specialist habitat that frequent burning is having a negative impact on the requirements, they were not caught in pitfall traps but environment. /Alan York only in leaf litter samples collected from near large logs on unburnt sites. They were not found on frequently burnt sites. In these forests they could be considered an uncommon species with high conservation status. /Alan York

198 Bushfire and forest invertebrates

3. RESULTS

The results for the various components of the with patterns quite variable and independent of project are reported separately, and then discussed treatment and but not spatial location (position) in terms of their relevance to the existing within the study area (see Figure 3.1B). experimental fire regime. 3.1.1.3 Mid-sized Shrub Layer The cover of Mid-sized Shrubs (50–100cm) on 3.1 ENVIRONMENTAL PARAMETERS sub-plots showed considerable variation overall The following sections summarise the results of (means ranged from 2–47%) with average an investigation into the effects of frequent (mean±s.e.) values for unburnt and burnt plots burning on environmental variables as they relate 14.3±0.8 and 14.5±0.8 respectively. Cover of Mid- to components of terrestrial invertebrate habitat. sized Shrubs was similarly variable (“patchy”) on Where differences are described as “significant”, burnt and unburnt plots. this refers to statistical significance at a probability Mean values of Mid-sized Shrub cover were level of 0.05. Where results were not considered not significantly different between burnt and to be statistically significant, the general nature of unburnt plots, however there were significant any observed patterns is described. “Box and spatial trends, with patterns quite variable and Whisker” plots are utilised to graphically independent of treatment and but not spatial represent variation in environmental variables. location (position) within the study area (see Figure The box represents the interquartile range 3.1C). (25th–75th percentile) with the median shown. 3.1.1.4 Tall Shrub Layer The whiskers indicate the range of values which The cover of Tall Shrubs (100–150cm) on sub- lie within 1.5 box lengths of the upper and lower plots showed moderate variation overall (means quartile (75th and 25th percentile respectively). ranged from 2–12%) with average (mean±s.e.) values for unburnt and burnt plots 5.5±0.4 and 3.1.1 Understorey Vegetation Structure 1.9±0.2 respectively. Cover of Tall Shrubs was less 3.1.1.1 Ground Herb Layer variable (“patchy”) on unburnt plots. The cover of Ground Herbs (0–20cm) on sub- Mean values of Tall Shrub cover were plots showed considerable variation overall significantly different between burnt and (means ranged from 29–98%) with average unburnt plots, with substantially lower cover on (mean±s.e.) values for unburnt and burnt plots frequently burnt plots. There were however 72.5±1.5 and 80.1±1.3 respectively. Cover of significant spatial trends, with patterns quite Ground Herbs was less variable (“patchy”) on variable and independent of treatment and but not burnt plots. spatial location (position) within the study area (see Mean values of Ground Herb cover were not Figure 3.1D). significantly different between burnt and 3.1.1.5 Very Tall Shrub Layer unburnt plots, however there were significant The cover of Very Tall Shrubs (150–200cm) on spatial trends, with patterns quite variable and sub-plots showed moderate variation overall independent of treatment and spatial location (means ranged from 0–10%) with average (position) within the study area (see Figure 3.1A). (mean±s.e.) values for unburnt and burnt plots 3.1.1.2 Small Shrub Layer 2.8±0.3 and 0.2±0.6 respectively. Cover of Very The cover of Small Shrubs (20–50cm) on sub- Tall Shrubs was less variable (“patchy”) on plots showed considerable variation overall unburnt plots. (means ranged from 12–91%) with average Mean values of Very Tall Shrub cover were (mean±s.e.) values for unburnt and burnt plots significantly different between burnt and unburnt 49.5±1.7 and 57.7±1.6 respectively. Cover of Small plots, with substantially lower cover on frequently Shrubs was less variable (“patchy”) on burnt plots. burnt plots. There were however significant spatial Mean values of Small Shrub cover were not trends, with patterns quite variable and significantly different between burnt and unburnt independent of treatment and spatial location plots, however there were significant spatial trends, (position) within the study area (see Figure 3.1E).

199 Australia’s Biodiveristy - Responses to Fire

Figure 3.1 Understorey Vegetation Structure

200 Bushfire and forest invertebrates

Figure 3.2 Effects of repeated burning upon Litter components.

201 Australia’s Biodiveristy - Responses to Fire

3.1.1.6 Summary 3.1.2.2 Bark The amount (cover) of vegetation in the first Mean biomass of Bark on sub-plots varied from metre above the ground (Ground Herbs, Small 0.50–2.52 t.ha.-1, with the mean (±s.e.) values for Shrubs & Mid-sized Shrubs) was not significantly unburnt and burnt plots 1.35 (±0.06) and 0.86 different between burnt and unburnt plots. (±0.06) t.ha.-1 respectively. Bark biomass was quite Significant interactions between the three factors variable by nature and considerably more variable (treatment, position & sub-plot) however reflects the (“patchy”) on burnt plots. spatially variable nature of these understorey Mean values of Bark biomass were vegetation components, irrespective of fire history significantly different between burnt and (treatment) and large- (position) & small-scale (sub- unburnt plots, with unburnt plots having (on plot) location, within the study area. Cover of average) higher Bark biomass. There were no Ground Herbs and Small Shrubs was however less obvious broad spatial trends with this variable (see spatially variable (“patchy”) on burnt plots, while Figure 3.2B), however there was substantial the cover of Mid-sized Shrubs was similarly within-plot variation, irrespective of treatment, with spatially variable on burnt and unburnt plots. primarily a greater level of spatial variability In contrast, the amount (cover) of vegetation within burnt plots compared to unburnt plots. in the second metre above the ground (Tall & 3.1.2.3 Twigs 0-6mm Very-tall Shrubs) was significantly different Values of mean Twig biomass (0–6mm diam.) on between burnt and unburnt plots. There was sub-plots varied from 0.99–5.22 t.ha.-1, with the substantially lower cover on frequently burnt mean (±s.e.) values for unburnt and burnt plots plots, and it was more spatially variable. There 3.65 (±0.01) and 1.84 (±0.08) t.ha.-1 respectively. were however significant interactions between the Twig biomass (0–6mm) was quite variable by three factors (treatment, position & sub-plot) again nature and slightly more variable (“patchy“) on reflecting the spatially variable nature of these burnt plots (see Figure 3.2C). understorey vegetation components, irrespective Mean values of Twig biomass were of fire history (treatment) and large- (position) & significantly different between burnt and small-scale (sub-plot) location, within the study unburnt plots, with unburnt plots having (on area. average) higher Twig biomass. There was however 3.1.2 Litter Biomass substantial within-plot variation, irrespective of treatment, reflecting the substantial spatial For purposes of analysis, Dry Litter Biomass was “patchiness” of this variable. considered in 5 categories compatible with other studies of “fine fuel”. These categories were: 3.1.2.4 Twigs 6–25mm Leaves, Bark, Twigs 0–6mm & 6–25mm diameter, Values of mean Twig biomass (6–25mm diam.) on -1 and Miscellaneous (very fine) Material. Green sub-plots varied from 0.35–6.19 t.ha. , with the (live) and cured (dead) vegetation data collected as mean (±s.e.) values for unburnt and burnt plots -1 part of the fuel studies were not used in this 2.72 (±0.21) and 1.53 (±0.16) t.ha. respectively. analysis. Twig biomass (0-6mm) was quite variable by nature and slightly more variable (“patchy”) on burnt 3.1.2.1 Leaves plots (see Figure 3.2D). Mean biomass of Leaves on sub-plots varied from Mean values of Twig biomass were -1 2.02–9.84 t.ha. , with the mean (±s.e.) values for significantly different between burnt and unburnt and burnt plots 7.54 (±0.19) and 4.23 unburnt plots, with unburnt plots having (on -1 (±0.14) t.ha. respectively. Biomass of Leaves was average) higher Twig biomass. There was however similarly variable (“patchy”) on burnt and unburnt substantial within-plot variation, particularly plots. within burnt plots. Mean values of biomass of Leaves were significantly different between burnt and 3.1.2.5 Very Fine (Miscellaneous) Material unburnt plots, with unburnt plots having (on Mean biomass of Miscellaneous Material on sub- -1 average) higher leaf biomass. There was an plots varied from 0.33–3.94 t.ha. , with the mean obvious spatial (N-S) trend with regard to this (±s.e.) values for unburnt and burnt plots 2.28 -1 variable, which was more apparent for unburnt (±0.11) and 0.85 (±0.04) t.ha. respectively. plots (see Figure 3.2A). Miscellaneous Material biomass was quite variable by nature but similarly variable (“patchy”) on burnt and unburnt plots.

202 Bushfire and forest invertebrates

Mean values of Miscellaneous Material were 3.1.3 Sticks & Logs significantly different between burnt and Sticks & Logs in excess of 2.5cm diameter (“heavy unburnt plots, with unburnt plots having (on fuel”) ranged overall from 2.5cm to 170cm average) higher amounts of Miscellaneous diameter, with a number of large logs still present Material. There were obvious broad spatial trends on the forest floor following the 1959 post- with this variable, particularly for burnt plots (see logging silvicultural treatment (T.S.I.). Figure 3.2E), however there was substantial There were 474 Sticks & Logs recorded within-plot variation, irrespective of treatment. overall on unburnt plots, ranging from 3.1.2.6 Summary 2.5–170cm diameter (mean±s.e.= 15.2±0.9cm) Twig (0–6mm & 6–25mm) and Bark biomass were with a similar pattern apparent on burnt plots significantly higher on unburnt plots, however (423 overall, ranging from 2.5–150cm diameter, there was substantial within-plot variation (spatial mean±s.e.= 13.6±0.9cm). Numbers of Sticks & “patchiness”) of these variables, irrespective of Logs per 40m transect on each sub-plot varied treatment. overall from 3–47, with average numbers slightly The biomass of Leaves was also significantly higher on unburnt sub-plots. higher on unburnt plots, with a spatial trend For purposes of analysis, size (diameter) of apparent, irrespective of treatment. This trend Sticks & Logs were divided into 5 size classes based was more apparent however on unburnt plots. on a visual inspection of the frequency distribution. Values for the very fine litter component Mean (±s.e.) numbers of Sticks & Logs in these 5 (Miscellaneous Material) were significantly higher size categories are shown in Table 3.1. on unburnt plots, with a spatial patterning Analysis of frequency distributions within apparent on burnt but not unburnt plots. nominated size categories revealed that there was Twig biomass (0–6mm & 6–25mm) was no obvious overall spatial pattern with regard to slightly more variable (“patchy”), and Bark the distribution of Sticks & Logs on either burnt biomass considerably more variable on burnt and unburnt plots, although considerable plots. In contrast, levels of within-plot variation variation was apparent, particularly for sticks and were similar for both the biomass of Leaves and of small logs between 10 and 50cm diameter. Field Miscellaneous Material on burnt and unburnt inspections revealed a change in character plots. however for larger logs on burnt plots with considerable charring and desiccation of the outer surface.

Table 3.1 Numbers of sticks & logs per sub-plot in 5 size classes (mean±s.e.) Diameter Unburnt (control) plots mean Frequently burnt plots mean Class 123456±se123456±se 2.5–10cm 9.8 6.3 10.3 12.5 14.0 12.5 10.9 9.0 8.3 9.0 16.3 8.9 11.0 10.3 ± 1.1 ± 2.1 ± 0.6 ± 3.0 ± 2.9 ± 1.1 ± 1.1 ± 3.1 ± 2.0 ± 1.5 ± 3.7 ± 1.7 ± 3.2 ± 1.1 10–25cm 6.5 4.0 7.3 5.3 5.8 6.3 5.9 5.0 3.0 5.3 10.5 3.8 4.3 5.3 ± 2.9 ± 2.0 ± 1.5 ± 1.7 ± 1.5 ± 3.2 ± 0.8 ± 2.4 ± 1.8 ± 1.1 ± 3.2 ± 0.9 ± 1.1 ± 0.9 25–50cm 2.8 1.0 1.0 2.3 1.5 2.5 1.8 1.5 0.8 1.8 1.8 2.0 0.5 1.4 ± 1.5 ± 0.7 ± 0.7 ± 1.1 ± 0.6 ± 1.6 ± 0.4 ± 0.5 ± 0.3 ± 1.4 ± 0.8 ± 1.1 ± 0.3 ± 0.3 50–75cm 1.0 1.0 0.8 0.3 0.8 0.3 0.7 0.8 0.3 0.5 0.3 0.3 0.5 0.4 ± 0.4 ± 0.4 ± 0.3 ± 0.3 ± 0.5 ± 0.3 ± 0.1 ± 0.5 ± 0.3 ± 0.3 ± 0.3 ± 0.3 ± 0.3 ± 0.1 75+ cm 0.5 0.5 0.0 1.0 0.8 0.3 0.5 0.5 0.5 0.3 0.0 0.0 0.5 0.3 ± 0.3 ± 0.5 ± 0.0 ± 0.7 ± 0.5 ± 0.3 ± 0.2 ± 0.5 ± 0.3 ± 0.3 ± 0.0 ± 0.0 ± 0.3 ± 0.1 Total 20.5 12.5 19.3 21.3 23.3 21.8 19.8 16.8 12.8 16.8 28.8 14.0 16.8 17.6 ± 5.2 ± 2.1 ± 2.8 ± 4.1 ± 3.7 ± 8.9 ± 1.9 ± 5.9 ± 3.0 ± 2.6 ± 5.1 ± 3.4 ± 4.7 ± 1.9

203 Australia’s Biodiveristy - Responses to Fire

3.1.4 Insolation 3.1.5.2 Top-soil Hardness Overall measures of the mean Top-soil Hardness Overall measures of the mean Insolation Index (shear) index varied from 1.9–10.0, with mean (Percentage of Light at Ground Level) varied from (±s.e.) values for unburnt and burnt plots 5.3 1.1 - 35.5%, with the mean (±s.e.) values for (±0.1) and 6.3 (0.2) respectively. Mean values of unburnt and burnt plots 7.3 (±0.7) and 16.4 (±1.1) Top-soil Hardness were not significantly different respectively. Insolation was extremely “patchy” by between burnt and unburnt plots. nature with the CV ranging from 31.5 - 68.6% There was a significant position effect (mean = 45.8%) on unburnt plots and 20.4 - reflecting a general north-south spatial pattern in 51.6% (mean = 30.6%) on burnt plots. Insolation soil hardness (see Figure 3.4B). There was was therefore less spatially variable (“patchy”) on however a slight tendency for a treatment/position unburnt plots. interaction caused by substantially higher values Mean values of Insolation were significantly on two burnt plots, 2 and 4 (positions 1 and 4). different between burnt and unburnt plots, with burnt plots having (on average) higher percentage light levels at ground level. There was however a general trend for increasing Insolation from north to south on burnt plots, with a different pattern evident on unburnt plots. This position effect was significant and is illustrated in Figure 3.3. 3.1.5 The Soil Environment

3.1.5.1 Top-soil Moisture Overall measures of mean Top-soil Moisture content on sub-plots varied from 4.5–17.7%, with mean (±s.e.) values for unburnt and burnt plots of 10.3% (±0.6) and 8.4% (±0.4) respectively. Mean values of Top-soil Moisture were significantly different between burnt and unburnt plots, with unburnt plots having (on average) higher percentage moisture levels. There was a significant large-scale spatial (position) effect, irrespective of treatment, with a slight increase apparent from position 1 to 3, and a gradual decline thereafter (see Figure 3.4A). Figure 3.4 Top-soil Moisture and Top-soil Hardness

3.2 INTER-RELATIONSHIPS BETWEEN ENVIRONMENTAL VARIABLES

The Principle Components Analysis (PCA) identified 8 significant themes in the environmental data. The 31 habitat variables were thus simplified to 8 new variables (Components) representing independent (uncorrelated) “patterns” in the environment. These Figure 3.3 Percentage Light at Ground Level on burnt Components are displayed in a matrix which and unburnt plots. shows the loadings of each variable for each Component (Table 3.2). These 8 Components described 75.1% of the overall variance in the environmental data set, thus substantially simplifying and clarifying structural patterns.

204 Bushfire and forest invertebrates .645 –.341 –.591 .529 .377 .721 .853 .847 .851 –.818 –.319 .373 .549 .638.861 .817 –.437 23.0 21.2 7.0 6.1 5.4 4.8 4.0 3.6 Logs 50+cm CV Topsoil hardness CV Topsoil CV Leaves tall shrubsCV Very –.376 Table 3.2 Significant Principle Components (eigenvalues > 1) showing variable loadings 0.3 Table VariableGround herbsCV Ground herbsSmall shrubsCVSmall shrubsMid-sized shrubsCV Mid-sized shrubsAspect Component 1 .946 –.949 Component 2 –.938 .940 .753 –.779 Component 3 .596 Component 4 Component 5 Component 6Component 7 Component 8 .367 SlopeEigenvalue explained % Variance Cumulative Variance 23.0 7.127 44.2 6.570 –.447 51.2 2.174 57.3 1.903 1.665 62.7 1.482 67.5 1.244 71.5 1.120 75.1% Twigs 0–6mmTwigs 6–10mmTwigs LeavesBarkCV BarkFine litter moistureTopsoil PLAG .311CV PLAG shrubsTall shrubsCV Tall tall shrubsVery –.438Sticks 0.5–10cm .614 Logs 10–25cm .901 Logs 25–50cm hardnessTopsoil –.349 –.550 .832 .567 CV Fine litter .686 .777CV twigs 0–6mm –.316 6–10mm CV Twigs .357 .614 –.684 .354 .352 .722 .378 .696 –.734 .339

205 Australia’s Biodiveristy - Responses to Fire

The first Component concerns the absence Table 3.3 Interpretation of environmental and spatial patchiness of Ground Herbs, Small Principal Components (based on identification Shrubs and Mid-sized Shrubs, and explains 23% of variables with high loadings - see Table 3.2). of the variance in the environmental data set (see Component Interpretation Table 3.3). This Component describes a pattern in the data whereby the first three structural 1 Absence and spatial patchiness of vegetation layers (0–1m) are positively correlated Ground Herbs, Small Shrubs and (ie. vary together). As the amount of vegetation in Mid-sized Shrubs. these layers increases, it becomes more spatially 2 Abundance of litter (Twigs, Leaves, homogenous. Conversely, low amounts of these Bark & Miscellaneous Material [very structural vegetation Components imply spatial fine litter]); high Top-soil Moisture; patchiness of the vegetation. There is a weak low and spatially variable amounts of tendency for there to be greater amounts of Insolation at ground level. vegetation in the first metre above the ground on 3 Presence and spatial homogeneity of sub-plots with more exposed (north-westerly) Tall Shrubs, presence of Very Tall aspects. Shrubs. The second Component concerns the 4 Presence of Sticks & Logs abundance of litter (Twigs, Bark, Leaves & (0.5–10cm, 10–25cm & 25–50cm Miscellaneous Material [very fine litter]), high diameter). Top-soil Moisture, and low and spatially patchy 5 Low and spatially variable Top-soil amounts of Insolation at ground level. This Hardness. describes a pattern that with increasing litter 6 Spatial patchiness of Leaves and levels (all Components), top-soil moisture levels Miscellaneous Material [very fine increase and the amount of insolation reaching litter]. ground level decreases and becomes more 7 Spatial patchiness of Twigs (0-6, spatially variable. This Component explains an 6–10mm) and Very Tall Shrubs. additional 21.2% of the variance in the 8 Presence of large logs (50+cm environmental data set. diameter) and low slope angles. The third Component concerns the presence and spatial homogeneity of Tall Shrubs, and presence of Very Tall Shrubs, and explains an As the first 3 Components together explain additional 7% of the variance in the environmental over half (51.2%) of the variance in the data set. This Component describes a pattern environmental data set (Table 3.3), their whereby when vegetation occurs in the upper Component scores were plotted to see how the understorey layers (1–2m) it tends to occur in both individual sub-plots were responding (Figure 3.5). elements of these layers, however it is spatially The complexity of 3-dimensions is displayed in patchy. two plots for clarity. The “elevation” view Subsequent Components do not explain illustrates the projection of Components 1 & 2, substantial additional variance in the representing 44.2% of the variance in environmental data set, however they do illustrate environmental variables. The “plan” view that many of the patterns in the habitat data are illustrates the projection of Components 1 & 3, independent (ie. uncorrelated). For example, the representing 23% and 7% of the variance number of Sticks & Logs in certain size categories respectively. (Component 4) are inter-correlated and unrelated A number of features are apparent. Firstly, in to the amount of Litter biomass at sub-plots the elevation view, there is a distinct separation of (Component 2). Top-soil Hardness (Component burnt and unburnt sub-plots along Component 5) is site-specific and independent of other 2. This reflects the distinct treatment effect environmental variables. The presence of large detected for Litter biomass (see 3.1.2). This logs on sites (Component 8) is also site-specific, separation is not apparent along Components 1 although there is a tendency for a greater number and 3, where there are no significant treatment on sub-plots with lower slope angles. effects. That is, the amount of vegetation in the first metre above ground is independent of treatment (see 3.1.1.6). Secondly, sub-plots within each plot are only loosely grouped together,

206 Bushfire and forest invertebrates

reflecting the lack of spatial patterning (position) in 3.3 TERRESTRIAL INVERTEBRATE understorey vegetation cover (see 3.1.1) and Litter COMMUNITIES Biomass (see 3.1.2). The relationship of these environmental In excess of 55,000 individuals from 24 broad patterns to the abundance and distribution of taxonomic groups were collected during pitfall terrestrial invertebrates is explored and reported trapping. These indicated a rich fauna with in the following sections of this report. representatives from the Chelicerata (spiders, ticks & mites, pseudoscorpions, harvestmen), Crustacea (landhoppers, slaters), Chilopoda (centipedes), Diplopoda (millipedes), and a large number of Insect Orders & Families (see Table 3.4). 3.3.1 Ordinal Diversity Of the 24 Orders recorded overall, ordinal diversity on sub-plots varied from 11–17, with mean (±s.e.) values for unburnt and burnt plots 14.0 (±0.3) and 12.7 (±0.2) respectively. A 2-way ANOVA indicated a significant effect (F1,5 = 20.65 P = 0.006) of treatment, but not position (F5,36 = 0.61 P = 0.693) on the number of Orders represented on sub-plots. Unburnt sub-plots were significantly more diverse at the Ordinal level than burnt plots, although the actual mean difference was slight (1 Order on average). Interaction between the factors treatment and position was not significant (F5,36 = 0.61 P = 0.693). This result suggests that frequent burning has slightly reduced the terrestrial invertebrate diversity at the Ordinal level (see Figure 3.6).

Figure 3.5 Habitat conditions on sub-plots as shown by scores on Components 1, 2 & 3.

Figure 3.6 Spatial Variation in Ordinal Diveristy

3.3.2 Invertebrate Abundance Numerically, the most abundant groups overall were the springtails (33.1%), ticks & mites (23.9%), ants (23.1%), bugs (4.2%), beetles (4.0%), bees & wasps (2.8%), insect larvae (2.7%), flies (2.6%) and spiders (2.2%), with these nine groups making up 98.6% of the total number of

207 Australia’s Biodiveristy - Responses to Fire iance results testing the ction 3.3.2). mean (s.e.) [n] mean (s.e.) [n] (millipedes) - - [2] - - - [3] [5] (centipedes) - - [6] - - - [7] [13] Diplopoda Araneae (spiders)Acarina (ticks & mites)Pseudoscorpionida (pseudoscorpions)Opilionida (harvestmen)CRUSTACEA Amphipoda (landhoppers)Isopoda (slaters) -UNIRAMIA Chilopoda 227.8 28.7Hexapoda Collembola (springtails) - -Diplura (diplurans) (20.3) -Blattodea (cockroaches) (1.5)Isoptera (termites) [5,468]Dermaptera (earwigs) [9]Orthoptera (grasshoppers, crickets) 6.4 -Embioptera (embiids) [688]Psocoptera (booklice) 352.2 -Hemiptera (bugs)Thysanoptera (thrips) 2.8 (1.0) [0] 0.7Neuroptera (lacewings) (21.6)Coleoptera (beetles) P,T - 1Diptera (flies) - 0.7 [153] (caddisflies)Tricoptera TP - (0.5) [8,452]Lepidoptera (moths, butterflies) (0.2)Hymenoptera (bees, wasps) - 328.9Hymenoptera (ants) - - (0.2) [67] - 61.9 2.1 [17] 21.5 - (28.7) - {T} 37.9 TP - [16] - [1]{} indicates not tested but trend apparent from data. - 29.0 - (7.1) (0.5) (1.5) - [7,895] 17.8 [7] (3.2) - 2.3 {T} 417.8 [1,487] [0] 10.3 {T} [13,363] - [50] (2.0) [0] [517] 416.2 (1.4) - [909] {T} (0.5) - (28.7) [0] (0.9) [4] [696] [12] (20.1) [1,205] 1.2 [427] 2.0 - [10,028] [2] T - [54] [248] [9,988] 2.4 - [13] (0.2) - [1] (0.3) P,T, - [1] [18,440] - [401] (0.4) - - [55] TP T,P, 35.0 T,TP [28] [49] - 1.2 [1] - 55.5 - - 119.5 (3.2) [58] - 42.3 - 34.9 [95] (0.4) (3.6) [66] - - (12.3) [7] - [841] (2.9) - [74] (2.7) [1,332] [0] - [29] [2,867] - [2,328] [1] [1,016] [8] [3] [839] [2,241] - [12,895] [7] [1] [79] [1,443] [9] [1,535] [1] [3] [0] [1] [21] [2] Table 3.4 Abundance of terrestrial invertebrates on frequently burnt and unburnt (control) plots, showing 3-way Analysis Var Table of treatment (burnt/unburnt) and position (N-S). effects TaxaCHELICERATA Burnt Plots * Anova#Larvae (insect)Totals* For Burnt and Unburnt replicates: values represent mean (± standard error) for n = 24, and [total caught] P = Position, TP indicates significant interaction term (refer Se # Anova results show significant effects where T = Treatment, Unburnt Plots * 24.2 Total (3.2) [580] [29,078] T 37.5 (3.1) [899] [1,479] [26,696] [55,774]

208 Bushfire and forest invertebrates

organisms caught. The first three groups were no significant position or interaction effects (springtails, ticks & mites, and ants) represented although values for burnt sub-plots at position 1 80% of individuals caught. were, on average, much lower (see Figure 3.7). The numbers collected from several groups were insufficient to comment about possible effects of frequent burning. These were the pseudoscorpions, harvestmen, centipedes, millipedes, diplurans, termites, embiids, booklice, lacewings, caddisflies, moths and butterflies. For these groups the trapping method used may not have been the most appropriate and has potentially contributed to the low capture rate. While the low numbers collected for several other taxa precluded statistical analysis, a general level of knowledge concerning their habitat requirements permits a few preliminary observations to be drawn. Frequent burning Figure 3.7 Abundance of Collembola on burnt and appears to have led to a reduction in the numbers unburnt plots of amphipods, cockroaches and earwigs, and an increase in the numbers of grasshoppers & crickets, and thrips. 3.3.2.3 Ticks & Mites For ten broad taxonomic groups there were Ticks and mites (Acarina) were numerically the sufficient data to permit statistical testing. A two- second most abundant organisms found during way Analysis of Variance procedure (ANOVA) was this trapping program, accounting for about 24% used to investigate treatment (burnt vs unburnt) of individuals. A 2-way Analysis of Variance and position (large spatial scale) effects, and (ANOVA) procedure (with ln(x+1) possible interactions between the factors. transformation) indicated that the effect of Significant treatment and position interaction treatment was significant (F1,5 = 21.03 P = 0.006), effects were detected for a number of taxa. with approximately 31% less individuals caught 3.3.2.1 Isopods (on average) on burnt plots. Variation in relative abundance between Overall only 401 slaters (Isopoda) were caught in plots within each treatment was high, with a pitfall traps. Although approximately 38% less significant position effect (F individuals were caught (on average) on burnt 1,36 = 1.11 P<0.001), plots, the effect of treatment was not significant reflecting a spatial pattern in abundance, and (F = 2.84 P = 0.153). Variation in relative greater numbers of individuals being caught on 1,5 unburnt abundance between plots within each treatment was plots at position 5 & 6, with no obvious high, with a significant treatment/position interaction treatment effect observed at other positions (see Figure 3.8). (F5,36 = 3.78 P = 0.007). This was due to greater numbers of individuals being caught on unburnt plots at position 1, 4, 5 & 6, and on the burnt plot at position 2. Only 4 species were detected (Fiona Lewis, Macquarie University pers. comm.) precluding further reliable analysis of these data. 3.3.2.2 Springtails Springtails (Collembola) were numerically the most abundant organisms found during this trapping program, accounting for about 33% of individuals. Approximately 15% less individuals were caught (on average) on burnt plots, however a 2-way Analysis of Variance (ANOVA) indicated a non-significant effect of treatment (F1,5 = 2.79 P = 0.156). Variation in relative abundance between Figure 3.8 Abundance of Acarina on burnt and unburnt plots within each treatment was low, and there plots

209 Australia’s Biodiveristy - Responses to Fire

At this stage taxonomic difficulties preclude 3.3.2.6 Hemiptera detailed examination at the species level, however Overall 2,328 individual bugs (Hemiptera) were the matter is currently being discussed with the trapped, which represented 2.2% of the total Australian Museum. number of organisms caught. Hemipteran 3.3.2.4 Hymenoptera (excluding ants) abundance on sub-plots varied from 9–138 As a component of the Hymenoptera, bees and individuals, with the distribution of values wasps are not effectively sampled by pitfall positively skewed (mean±s.e.= 48.5±4.3, median = trapping. However, over 1,500 individuals were 37.5). The abundance of Hemiptera overall was caught, with no significant effects of treatment or more variable on burnt (23–138) compared to position. Values on unburnt sub-plots at position 2 unburnt (9-88) sub-plots. were substantially higher than at burnt sub-plots, Of the total number of Hemiptera caught however they were within the range experienced 841 (36%) were trapped on unburnt sub-plots by other plots from that treatment (see Figure 3.9). and 1,487 (64%) on burnt sub-plots. Mean (±s.e.) & median values for unburnt and burnt plots were 35.0 (±3.2) & 30.5 and 61.9 (±7.1) & 46.5 respectively. A 2-way Analysis of Variance (ANOVA) procedure (with ln(x+1) transformation) indicated that the effect of treatment was significant (F1,5 = 10.93 P=0.021) with approximately 77% more individuals caught (on average) on burnt plots (Figure 3.11). Variation in relative abundance between plots within each treatment was high, although the effect of position was not significant (F5,36 = 2.21 P = 0.075). Abundance was higher on all burnt plots except position 6, although interaction between Figure 3.9 Abundance of Hymenoptera (non-ants) on treatment and position was not significant (F = burnt and unburnt plots 5,36 2.91 P = 0.241). 3.3.2.5 Insect Larvae For insect larvae, a 2-way Analysis of Variance (ANOVA) procedure indicated that the effect of treatment was significant (F1,5 = 14.67 P = 0.012), with approximately 35% less individuals caught (on average) on burnt plots. Variation in relative abundance between plots within each treatment was low with no significant position or interaction effects (see Figure 3.10).

Figure 3.11 Abundance of Hemiptera on burnt and unburnt plots

3.3.2.7 Diptera Although pitfall trapping may not appear to be the preferred method of sampling flies (Diptera), 1,443 individuals were caught, representing 2.6% of the total number of organisms. Dipteran abundance on sub-plots varied from 8–64 individuals, with the distribution of values slightly Figure 3.10 Abundance of Insect Larvae on burnt and positively skewed (mean = 30.1±2.4, median = unburnt plots

210 Bushfire and forest invertebrates

28.0). The abundance of Diptera overall was more 21.5 (±1.5) & 22.0 and 28.7±1.5 & 28.5 variable on unburnt (9–64) compared with burnt respectively. There were (on average) 33% more (8–34) sub-plots. spiders on frequently burnt sub-plots. Of the total number of Diptera caught 1,016 A 2-way ANOVA indicated a non-significant (70%) were trapped on unburnt sub-plots and effect of treatment (F1,5 = 1.48 P = 0.278) and 427 (30%) on burnt sub-plots. Mean (±s.e.) & position (F5,36 = 1.45 P = 0.229), with a significant median values for unburnt and burnt plots were interaction detected between these two factors 42.3 (±2.9) & 43.0 and 17.8 (±1.4) & 16.0 (F5,36 = 3.24 P = 0.016). respectively. Mean values of spider abundance were A 2-way Analysis of Variance (ANOVA) therefore not significantly different between procedure indicated that the effect of treatment burnt and unburnt plots, although burnt plots was significant (F1,5 = 24.30 P = 0.004), with have (on average) higher numbers of individuals approximately 58% less individuals caught (on (see Figure 3.13). The significant interaction average) on burnt plots. between treatment and position reflects the reversal Variation in relative abundance between in the pattern of the treatment effect at position 3 plots within each treatment was high, with a and 4, where burnt plots had slightly lower spider marginally non-significant position effect (F5,36 = abundance than unburnt, and at position 6 where 2.22 P = 0.073) and a significant treatment/position the mean value on burnt plots (11.3±1.7) is interaction effect (F5,36 = 3.49 P = 0.011). This substantially higher than on unburnt plots was due to greater numbers of individuals being (4.8±0.5). caught on unburnt plots at position 2, 3, 4, 5 & 6, and on the burnt plot at position 1 (see Figure 3.12).

Figure 3.13 Abundance of Spiders on burnt and unburnt plots

Figure 3.12 Abundance of Diptera on burnt and unburnt plots 3.3.2.9 Beetles Overall 2,145 individual beetles were trapped, 3.3.2.8 Spiders which represented 4.0% of the total number of Overall 1,205 individual spiders (Araneae) were organisms caught. Beetle abundance on sub-plots trapped, which represented 2.2% of the total varied from 13–88 individuals, with the with the number of organisms caught. Spider abundance on distribution of values slightly positively skewed sub-plots varied from 8–42 individuals, with the (mean = 44.7, median = 43.5). The abundance of distribution of values slightly positively skewed beetles overall was slightly less variable on burnt (mean ±s.e= 25.1±1.1, median = 23.0). The (13–67) compared with unburnt (29–88) sub-plots. abundance of spiders overall was slightly more Of the total number of beetles caught, 1,269 variable on unburnt (8-40) compared with burnt (59%) were trapped on unburnt sub-plots and (16–42) sub-plots. 876 (41%) on burnt sub-plots. Mean (±s.e.) & Of the total number of spiders caught 515 median values for unburnt and burnt plots were (43%) were trapped on unburnt sub-plots and 52.9 (±3.2) & 47.5 and 36.5 (±2.9) & 34.0 688 (57%) on burnt sub-plots. Mean (±s.e.) & respectively. There were (on average) 31% less median values for unburnt and burnt plots were beetles on frequently burnt sub-plots.

211 Australia’s Biodiveristy - Responses to Fire

A 2-way ANOVA (with logex position (F5,36 = 12.01 P<0.001), with a significant transformation) indicated a significant effect of interaction detected between these two factors treatment (F1,5 = 8.75 P = 0.032) and position (F5,36 = 2.82 P = 0.030). (F5,36 = 6.65 P<0.001), with a significant Mean values of ant abundance were interaction detected between these two factors therefore significantly different between burnt (F5,36 = 3.44 P = 0.012). and unburnt plots, with burnt plots having (on Mean values of beetle abundance were average) much higher numbers of individuals. The therefore significantly different between burnt position effect and treatment/position interaction and unburnt plots, with burnt plots having (on reflects the strong spatial trend in ant abundance average) lower numbers of individuals. The with higher values at position 1, particularly for position effect reflects the strong spatial trend in burnt plots (see Figure 3.15). beetle abundance, particularly for burnt plots (see Figure 3.14). The significant interaction effect reflects the reversal in the pattern of the treatment effect at position 4, where burnt plots had slightly higher beetle abundance than unburnt plots.

Figure 3.15 Abundance of Ants on burnt and unburnt plots

3.3.2.11 Summary This forest environment has a abundant and Figure 3.14 Abundance of Beetles on burnt and unburnt plots diverse terrestrial invertebrate fauna with in excess of 55,000 individuals from 24 broad taxonomic groups collected during a one week sampling 3.3.2.10 Ants period using pitfall traps. Numerically, the most Ants (Hymenoptera:Formicidae) represented the abundant groups overall were the springtails third most abundant group trapped (12,895 (33.1%), ticks & mites (23.9%), ants (23.1%), bugs individuals), accounting for 23% of the total (4.2%), beetles (4.0%), bees & wasps (2.8%), insect catch. Ant abundance on sub-plots varied from larvae (2.7%), flies (2.6%) and spiders (2.2%). 24–778 individuals, with the with the distribution Due to their low numbers, it was not of values slightly positively skewed (mean±s.e.= possible to comment on the effects of frequent 268.6±26.7, median = 219.0). The abundance of burning for: pseudoscorpions, harvestmen, ants overall was considerably more variable on centipedes, millipedes, diplurans, termites, burnt (155–778) compared with unburnt embiids, booklice, lacewings, caddisflies, moths (24–300) sub-plots. and butterflies. For these groups the trapping Of the total number of ants caught 2,867 method used may not have been the most (22%) were trapped on unburnt sub-plots and appropriate and has potentially contributed to the 10,028 (78%) on burnt sub-plots. Mean (±s.e.) & low capture rate. While the low numbers median values for unburnt and burnt plots were collected for several other taxa precluded 119.5 (±12.3) & 102.5 and 417.8 (±28.7) & 401.5 statistical analysis, frequent burning appears to respectively. There were (on average) 250% more have led to a reduction in the numbers of ants on frequently burnt sub-plots. amphipods, cockroaches and earwigs, and an A 2-way ANOVA indicated a significant increase in the numbers of grasshoppers & effect of treatment (F1,5 = 77.82 P < 0.001) and crickets, and thrips.

212 Bushfire and forest invertebrates

Overall, the number of invertebrate Orders The groups studied at morphospecies level had been significantly reduced on sub-plots were the: experiencing frequent burning, although the Hemiptera (bugs): magnitude of this decrease was small (average ≈ 1 as most are terrestrial and phytophagous Order). (plant-feeding), they are a group which have For ten broad taxonomic groups there were a close association with plant communities. sufficient data to permit statistical testing using Diptera (flies): the Analysis of Variance (ANOVA) procedure to although highly mobile as adults, the flies investigate the effects of frequent burning have particular requirements with regard to (treatment) and patterns due to large-scale spatial larval food sources; usually moist, decaying effects (position). These results are summarised in plant and animal material. Many species are Table 3.5 and indicate a variety of responses to parasitic on the larvae of other insect orders. frequent burning. Seven groups (isopods, Araneae (spiders): springtails, ticks & mites, bees & wasps, insect spiders are a major group of predators in larvae, flies and beetles) showed substantial forest ecosystems exploiting a variety of decreases in abundance following frequent habitats. They live in burrows or crevices in burning. These decreases ranged from 15 to 58%, the ground, amongst leaf litter or in but were only statistically significant for ticks & vegetation. mites, insect larvae, flies and beetles. High spatial Coleoptera (beetles): variability in abundance for isopods, springtails, beetles utilise a diverse range of habitats & and bees & wasps possibly contributed to the lack micro- habitats, with a variety of feeding of statistical significance. strategies (adults include herbivores, Three groups showed substantial increases predators & scavengers, while larval forms in abundance following frequent burning. These feed either internally or externally on plants were statistically significant for bugs (77%) and ants and fungal products). (250%), but not for spiders (33%). Both spiders Formicidae (ants): and ants showed considerable spatial variability in ants are one of the most numerous and their numbers. widespread groups in Australian ecosystems. They have a diverse diet, and utilise a variety 3.3.3 Invertebrate Species Richness of feeding strategies from predators and Five invertebrate groups were identified to scavengers, to plant eaters and fungus morphospecies (see Section 2.4) in order to feeders, with frequent and varied further investigate the impact of repeated burning interactions with other invertebrate groups. on species richness, and the related aspects of These groups are discussed separately with community composition and structure. These related issues considered in Sections 3.3.3.6 groups utilise a diversity of micro-habitats and & 7. niches and are representative of the range of terrestrial invertebrates found in these forest environments. Table 3.5 Changes in mean abundance following frequent burning for selected terrestrial invertebrate taxa. Change with Statistically Large-scale Taxa Common name frequent burning significant ? (P<0.05) spatial patterns ? Isopoda slaters, pill-bugs ⇓ 38% no yes Collembola springtails ⇓ 15% no no Acarina ticks & mites ⇓ 31% yes yes Hymenoptera (excl. ants) bees & wasps ⇓ 17% no yes Insecta insect larvae ⇓ 35% yes no Hemiptera bugs ⇑ 77% yes no Diptera flies ⇓ 58% yes no* Araneae spiders ⇑ 33% no yes* Coleoptera beetles ⇓ 31% yes yes* Formicidae ants ⇑ 250% yes yes* * indicates a statistically significant interaction between treatment and position effects.

213 Australia’s Biodiveristy - Responses to Fire

3.3.3.1 Hemiptera A 2-way ANOVA indicated a significant Overall, 44 morphospecies of Hemiptera were effect of treatment (F1,5 = 14.84 P = 0.012) and collected, 25 on unburnt plots and 26 on frequently position (F5,36 = 24.67 P = 0.031), with no burnt plots. Hemipteran richness on sub-plots significant interaction detected between these two varied from 1–8, with a normal distribution of factors (F5,36 = 1.82 P = 0.134). values (mean = 3.4 (±0.2), median = 3.0). Mean values of fly richness were therefore Mean (±s.e.) & median values for unburnt significantly different between burnt and and burnt plots were 3.2 (±0.3) & 3.0 and 3.7 unburnt plots, with burnt plots having (on (±0.3) & 3.0 respectively. There were (on average) average) lower numbers of morphospecies (see 16% more Hemipteran species on frequently Figure 3.17). The weak position effect identified burnt sub-plots. The number of morphospecies for fly abundance (see 3.3.2.7) was more evident was similarly variable on unburnt (1–8) and for fly species richness, but only for burnt plots. burnt (1–7) sub-plots. Values of species richness were quite consistent A 2-way ANOVA (with logex between unburnt plots. transformation) indicated a non-significant effect of treatment (F1,5 = 1.99 P = 0.217) and position (F5,36 = 1.49 P = 0.218), with a non-significant interaction detected between these two factors (F5,36 = 0.60 P = 0.699). Mean values of Hemipteran richness were therefore not significantly different between burnt and unburnt plots, with both treatments having (on average) similar numbers of morphospecies (see Figure 3.16).

Figure 3.17 Richness of Flies on burnt & unburnt plots

3.3.3.3 Spiders Overall, 63 morphospecies of spiders were collected, 32 on unburnt plots and 47 on frequently burnt plots. Spider richness on sub- plots varied from 1–10, with a normal distribution of values (mean = 4.9 (±0.3), median = 5.0). Mean (±s.e.) & median values for unburnt and Figure 3.16 Richness of Hemiptera on burnt & unburnt burnt plots were 4.4 (±0.4) & 4.0 and 5.6 (±0.4) & plots 6.0 respectively. There were (on average) 27% more spider species on frequently burnt sub-plots. The 3.3.3.2 Diptera number of morphospecies was similarly variable on Overall, 77 morphospecies of flies were collected, 66 unburnt (1–8) and burnt (2–10) sub-plots. on unburnt plots and 46 on frequently burnt plots. A 2-way ANOVA indicated a non-significant Fly richness on sub-plots varied from 0 –16, with a effect of treatment (F1,5 = 2.43 P = 0.180) and normal distribution of values (mean = 7.9 (±0.6), position (F5,36 = 0.92 P = 0.482), with a weak median = 8.0). interaction detected between these two factors Mean (±s.e.) & median values for unburnt (F5,36 = 2.27 P = 0.069). and burnt plots were 10.1 (±0.6) & 10.0 and 5.7 Mean values of spider richness were (±0.8) & 6.0 respectively. There were (on average) therefore not significantly different between 44% fewer fly species on frequently burnt sub- burnt and unburnt plots, although burnt plots plots. The number of morphospecies was similarly have (on average) slightly higher numbers of variable on unburnt (6–16) and burnt (0–12) sub- morphospecies (see Figure 3.18). The weak plots. position/treatment interaction effect identified for

214 Bushfire and forest invertebrates

spider abundance (see 3.3.2.8) was also evident for Mean values of beetle richness were spider species richness, largely due to higher therefore significantly different between burnt richness on burnt sub-plots at position 6. and unburnt plots, with burnt plots having (on average) lower numbers of morphospecies (see Figure 3.19). The strong position effect identified for beetle abundance (see 3.3.2.9) was not evident for beetle species richness, although there is a slight N-S decline evident on unburnt plots. The significant interaction reflects the reversal in the pattern of the treatment effect at position 3, where burnt sub-plots had slightly higher species richness (see Figure 3.19). 3.3.3.5 Ants Overall, 88 morphospecies of ants were collected, 70 on unburnt plots and 68 on frequently burnt Figure 3.18 Richness of Spiders on burnt and unburnt plots. Ant richness on sub-plots varied from plots 11–27, with a normal distribution of values (mean = 19.7 (±0.6), median = 19.5). Mean (±s.e.) & median values for unburnt 3.3.3.4 Beetles and burnt plots were 17.4 (±0.6) & 18.0 and 22.0 Overall, 139 morphospecies of beetles were (±0.6) & 22.0 respectively. There were (on collected, 86 on unburnt plots and 92 on average) 26% more ant species on frequently frequently burnt plots. Beetle richness on sub- burnt sub-plots. The number of morphospecies plots varied from 5–20, with a normal distribution was similarly variable on unburnt (11–24) and of values (mean = 11.3 (±0.5), median = 11.0). burnt (17–27) sub-plots. Mean (±s.e.) & median values for unburnt A 2-way ANOVA indicated a significant and burnt plots were 13.1 (±0.6) & 13.0 and 9.5 effect of treatment (F1,5 = 11.22 P = 0.020) but not (±0.5) & 9.0 respectively. There were (on average) position (F5,36 = 0.52 P = 0.757), with a significant 27% fewer beetle species on frequently burnt interaction detected between these two factors sub-plots. The number of morphospecies was (F5,36 = 2.88 P = 0.028). similarly variable on unburnt (8–20) and burnt Mean values of ant richness were therefore (5–15) sub-plots. significantly different between burnt and A 2-way ANOVA indicated a significant unburnt plots, with burnt plots having (on effect of treatment (F1,5 = 8.81 P = 0.031) but not average) higher numbers of morphospecies (see position (F5,36 = 2.24 P = 0.071), with a significant Figure 3.20). The strong position effect identified interaction detected between these two factors for ant abundance (see 3.3.2.10) was not evident (F5,36 = 2.86 P = 0.028). for ant species richness. The significant interaction reflects a lack of treatment effect at positions 1 and 4. 3.3.3.6 Scale Effects It is also important to recognise the effect that sampling intensity (scale) has on the detection of treatment effects for different taxa. With the bugs, results were consistent across a range of scales of measurement. The magnitude and direction of differences between unburnt and burnt species richness results for sub-plot (3.2 vs 3.7) and plot (7.5 vs 7.7) means and treatment totals (25 vs 26) were similar at these three scales (see Table 3.6). Similar patterns were apparent for flies, with a Figure 3.19 Richness of Beetles on burnt and unburnt similar magnitude of difference detected at the plots scale of sub-plot (10.1 vs 5.7), plot (25.8 vs 16.5) and treatment (66 vs 46).

215 Australia’s Biodiveristy - Responses to Fire

For ants, the magnitude of the difference in species richness detected between unburnt and burnt areas at the scale of sub-plot (17.4 vs 22.0) and plot (33.3 vs 38.7) were similar, although the magnitude of the difference was reduced at the scale of plot (compared with other taxa). The direction of the difference was however substantially reversed at the scale of treatment (70 vs 68). This would suggest a different situation to that with the spiders and beetles, with the species’ assemblages on burnt plots more similar (less diverse) than those on unburnt plots. Diversity Figure 3.20 Richness of Ants on burnt and unburnt plots on sub-plots within both unburnt and burnt plots would appear to be less similar than with For spiders however, while the magnitude other taxa, suggesting the differences lie at less and direction of species richness at the scale of than the scale of plot ( 1 hectare). These patterns sub-plot (4.4 vs 5.6) and plot (12.5 vs 15.8) were will be further explored in Section 3.3.4.5. similar, considerably more species were found 3.3.3.7 Summary overall on burnt compared to unburnt plots (47 Overall, 411 morphospecies were identified from vs 32). This would suggest that the species’ the five groups studied in detail. The beetles were assemblages on burnt plots are less similar (more the most species rich (139 species), followed by diverse) than those on unburnt plots, resulting in the ants (88), flies (77), spiders (63), and bugs (44). higher (between-habitat) diversity on frequently The results of analyses (ANOVA) burnt plots compared to unburnt plots. Diversity investigating the effects of frequent burning on sub-plots within both unburnt and burnt (treatment) and patterns due to large-scale spatial plots would appear to be similar, suggesting the effects (position) are summarised in Table 3.7, and differences lie at the scale of plot ( 1 hectare). indicate a variety of responses to frequent These patterns will be further explored in Section burning. Two groups, flies and beetles, 3.3.4.3. experienced a significant reduction in species For beetles, the magnitude of the difference richness on sub-plots following frequent burning detected between unburnt and burnt at the scale (44% and 27% reduction respectively). Two of sub-plot (13.1 vs 9.5) and plot (29.3 vs 27.0) were groups, the bugs and the spiders, showed an similar, however the direction was reversed at the increase in species richness on sub-plots (16% and scale of treatment (86 vs 92). This would suggest a 27% respectively), although these results were not similar situation as to that with the spiders, where statistically significant. The ants experienced a the species’ assemblages on burnt plots are less significant increase in sub-plot richness (26%) similar (more diverse) than those on unburnt plots. following repeated burning. Diversity on sub-plots within both unburnt and It was apparent that estimates of species burnt plots would appear to be similar, suggesting richness were influenced by the spatial scale of the differences lie at the scale of plot ( 1 hectare). measurement, with associated implications for the These patterns will be further explored in Section interpretation of observed treatment effects for the 3.3.4.4.

Table 3.6 Comparison of estimates of species richness at different scales of measurement. Taxa sub-plot (mean±s.e.) plot (mean±s.e.) treatment total unburnt burnt unburnt burnt unburnt burnt Bugs 3.2±0.3 3.7±0.3 7.5±0.5 7.7±0.9 25 26 Flies 10.1±0.6 5.7±0.8 25.8±0.6 16.5±3.6 66 46 Spiders 4.4±0.4 5.6±0.4 12.5±1.5 15.8±1.7 32 47 Beetles 13.1±0.6 9.5±0.5 29.3±0.8 27.0±3.1 86 92 Ants 17.4±0.6 22.0±0.6 33.3±1.3 38.7±1.1 70 68

216 Bushfire and forest invertebrates

different taxa. For bugs and flies results were These spatial patterns in estimates of species consistent across a range of scales of richness are a consequence of the composition of measurement, with the magnitude and direction invertebrate assemblages (communities) at the of differences between unburnt and burnt results varying scales of investigation. The nature of for sub-plot, plot and treatment similar. these patterns, and their interaction with For spiders, while the magnitude and environmental variables, will be further explored direction of species richness at the scale of sub- in Section 3.3.4. plot and plot were similar, considerably more 3.3.4 Community Composition species were found overall on burnt compared to unburnt plots. This suggested that species’ 3.3.4.1 Hemiptera assemblages on burnt plots were more diverse The 44 Hemipteran morphospecies were than those on unburnt plots, resulting in higher representative of 14 families, the infra-order (between-habitat) diversity. Diversity on sub-plots Dipsocoromorpha (not readily discernible to within both unburnt and burnt plots would family) and an unidentifiable Homopteran (see appear to be similar, suggesting the differences lie Table 3.8). The most diverse groups overall were at the scale of plot ( 1 hectare). the family Cicadellidae and the infra-order For beetles, the magnitude of the difference Dipsocoromorpha, containing 27% and 25% of detected between unburnt and burnt at the scale overall Hemipteran morphospecies respectively. of sub-plot and plot were similar, however the All three sub-orders were represented: the direction was reversed at the scale of treatment. Sternorrhyncha (5 morphospecies), the This would suggest a similar situation as to that Auchenorrhyncha (13 morphospecies) and the with the spiders, where the species’ assemblages Heteroptera (26 morphospecies). The on burnt plots are more diverse than those on Sternorrhyncha are mainly sedentary, often living unburnt plots. Diversity on sub-plots within both under waxy secretions or inside galls induced in the unburnt and burnt plots would appear to be host plants. Four morphospecies from this sub- similar, suggesting the differences lie at the scale order were found on unburnt plots and three on of plot (1 hectare). burnt plots (see Table 3.8). The Auchenorrhyncha For ants, the magnitude of the difference in (leaf- and plant-hoppers) are all plant-sap feeders as species richness detected between unburnt and adults and generally spend the bulk of their time on burnt areas at the scale of sub-plot and plot were plant foliage. Seven morphospecies from this sub- similar, although the magnitude of the difference order were found on unburnt plots and nine on was reduced at the scale of plot (compared with burnt plots. Most of the “true bugs”, the other taxa). The direction of the difference was Heteroptera, are plant-sap feeders, although some however substantially reversed at the scale of groups, such as the Reduviidae, are predatory. treatment, suggesting a different situation to that Thirteen morphospecies from this sub-order were with the spiders and beetles, with the species’ found on unburnt plots and fourteen on burnt assemblages on burnt plots less diverse. Diversity plots. Overall, a similar number of morphospecies on sub-plots within both unburnt and burnt plots were found on unburnt and burnt plots (24 and 26 would appear to be less similar than with other respectively) with similar proportions from the taxa, suggesting the differences lie at less than the three sub-orders. scale of plot ( 1 hectare). Although the two treatments are similarly diverse, both at the sub-ordinal and

Table 3.7 Changes in mean species richness following frequent burning for selected terrestrial invertebrate taxa. Taxa Common name Change with Statistically Large-scale frequent burning significant ? (P<0.05) spatial patterns ? Hemiptera bugs ⇑ 16% no no Diptera flies ⇓ 44% yes yes Araneae spiders ⇑ 27% no no* Coleoptera beetles ⇓ 27% yes no* Formicidae ants ⇑ 26% yes no* * indicates a statistically significant interaction between treatment and position effects.

217 Australia’s Biodiveristy - Responses to Fire

Table 3.8 Breakdown of Hemipteran morphospecies by sub-order and family. # Morphospecies Sub-order Family Common name unburnt burnt total Sternorrhyncha Coccidae scale insects 3 3 4 Homoptera* - 1 0 1 Auchenorrhyncha Cicadellidae leaf-hoppers 6 8 12 Fulgoroidea plant-hoppers 1 1 1 Heteroptera Dipsocoromorpha* - 8 4 11 Enicocephalidae - 1 0 1 Nabidae - 1 0 1 Tingidae lace bugs 1 0 1 Pentatomidae shield bugs 1 0 1 Lygaeidae seed bugs 1 1 2 Anthocoridae flower bugs 0 1 1 Miridae - 0 1 1 Thaumastocoridae - 0 1 1 Plataspidae - 0 1 1 Reduviidae assassin bugs 0 5 5 Totals 24 26 44 * Not readily discernible to Family level morphospecies level (see Section 3.3.3.1), only 7 sub-plot of a treatment, it is difficult to identify morphospecies (16%) were common to both clear patterns from these tabulated data (at the treatments (Group A - Table 3.9). When the species level). At this stage it is apparent however morphospecies are arranged to reflect their that unburnt plots have more species from the distribution across plots and treatments, it is clear infra-order Dipsocoromorpha (7 vs 3), while that there is a group of species (18) found only on burnt plots have greater numbers of species from unburnt sub-plots (41% - Group B) and a the family Reduviidae (5 vs 0). differenct group of species (19) found only on Although this table of relative abundance burnt sub-plots (43% - Group C). Frequent enables broad “assemblages” of species with burning would therefore appear to have resulted similar responses to disturbance to be identified, in the loss of up to 18 species of Hemipteran, these data are more clearly displayed in the form however the changed environment supports upto of a “bi-plot” derived from the CCA ordination 19 new species not found in unburnt areas. (see 2.5.3.3). This graphical display (Figure 3.21) Morphospecies found on both treatments shows the configuration of the environmental are representative of the families Cicadellidae variables, the scatter of sub-plots, and the (3 species), Coccidae (2 species), Fulgoroidea relationship between the two, giving an overview (1 species) and the infra-order Dipsocoromorpha of how community composition varies with the (1 species). Morphospecies found only on environment (Ter Braak 1986). unburnt sub-plots are representative of the A number of features are apparent from this families Cicadellidae (4 species), Coccidae bi-plot. Firstly, the minimal overlap of unburnt and (1 species), Nabidae (1 species), Enicocephalidae burnt sub-plots in ordination space reflects the (1 species), Lygaeidae (1 species), Pentatomidae largely dissimilar species assemblages of these two (1 species), Tingidae (1 species), the infra-order treatments. Secondly, the tighter clustering of burnt Dipsocoromorpha (7 species) and an unnamed sub-plots indicates a lower within-treatment Homopteran species. Morphospecies found only diversity compared to unburnt sub-plots (ie. a on burnt sub-plots are representative of the lower β-diversity). Richness values on burnt sub- families Cicadellidae (5 species), Coccidae plots were on average 16% higher than on unburnt (1 species), Miridae (1 species), Reduviidae sub-plots (see Section 3.3.3.1), however the high (5 species), Lygaeidae (1 species), Plataspidae similarity of sub-plots for this treatment mean that (1 species), Thaumastocoridae (1 species), the overall richness of unburnt and burnt plots was Anthocoridae (1 species) and the infra-order similar (25 vs 26 species). Unburnt sub-plots have Dipsocoromorpha (3 species). lower richness (α-diversity) but are less similar, As many of these families are represented resulting in higher “turnover” between sub-plots only by single individuals or found only on one (higher β-diversity). 218 Bushfire and forest invertebrates Table 3.9 Presence of Hemipteran morphospecies on burnt and unburnt plots Table

219 Australia’s Biodiveristy - Responses to Fire

The third feature concerns the contribution Brachycera (59 morphospecies). Nematocera adults of environmental variables to the differences in are generally slender with long legs, and have species composition for the two treatments. In the aquatic larvae (eg. mosquitoes, midges & sandflies) bi-plot (Figure 3.21) the length of the arrow or are gall makers. Twenty-five morphospecies from signifies the relative contribution of that variable this sub-order were found on unburnt plots and to species composition, and the direction signifies eighteen on burnt plots (see Table 3.10). its contribution to the differences between Brachycera adults are heavier set with relatively treatments. Unburnt sub-plots are characterised short legs (eg. House flies and March flies) and have by high levels of litter and high cover of tall and mainly terrestrial larvae (often found in damp soil very tall shrubs. Due to the correlation between and rotting vegetation). Forty-one morphospecies variables (see Section 3.2) these plots are also from this sub-order (Brachycera) were found on characterised by high top-soil moisture levels and unburnt plots and twenty-eight on burnt plots. low and spatially variable amounts of insolation at Overall, 30% less morphospecies were found on ground level. These environmental variables make burnt plots compared with unburnt plots (46 and the greatest contribution to the differences in 66 respectively) with similar trends for both sub- species composition between unburnt and burnt orders. plots. Burnt sub-plots are characterised by high The unburnt treatment was considerably levels of insolation at ground level, and to a lesser more diverse than the burnt treatment at the extent, steeper slopes, greater top-soil hardness morphospecies level (see Section 3.3.3.2) however 35 and greater cover of the herb & shrub component morphospecies (45%) were common to both of the understorey vegetation. Other treatments (Group A - Table 3.11). When the environmental variables make only a minor morphospecies are arranged to reflect their contribution to the observed differences in distribution across plots and treatments, it is clear Hemipteran species composition between burnt that there is a group of species (31) found only on and unburnt treatments. unburnt sub-plots (40% - Group B) and a different group of species (11) found only on burnt sub-plots (14% - Group C). Frequent burning would therefore appear to have resulted in the loss of up to 31 species of Diptera, however the changed environment supports up to 11 new species not found in unburnt areas. Morphospecies found on both treatments are representative of the families Phoridae (12 species), Cecidomyiidae and Sciaridae (5 species each), Chloropidae (3 species), Tachydromiinae, Ceratopogonidae, Chironomidae, Muscidae (2 species each), Dolichopodidae and Scatopsidae (1 species each). Morphospecies found only on Figure 3.21 Bi-plot from CCA ordination of Hemipteran unburnt sub-plots are representative of the families presence/absence data. (Ellipses represent 1 standard Tachydromiinae and Phoridae (4 species each), deviation unit around treatment centroids) Ceratopogonidae and Sciaridae (3 species each), Chloropidae, Drosophilidae, Muscidae, 3.3.4.2 Diptera Sphaeroceridae and Tachinidae (2 species each), Syrphidaea, Therevidae, Micropezidae, Tipulidae, The 77 Dipteran (fly) morphospecies were Cecodomyiidae, Mycetophilidae and Chironomidae representative of 2 sub-orders and 20 families (see (1 species each). Morphospecies found only on Table 3.10). The most diverse groups overall were burnt sub-plots are representative of the families the families Phoridae and Sciaridae, containing Ceratopogonidae (2 species), Tachinidae, 22% and 10% of morphospecies respectively. The Piophilidae, Calliphoridae, Cecidomyiidae, families Cecodomyiidae, Ceratopogonidae and Chloropidae, Muscidae, Tachydromiinae and Empididae each contained 9% of overall Sphaeroceridae (1 species each). morphospecies. As many of these families are represented Both Australian sub-orders were represented: only by single individuals or found only on one the Nematocera (28 morphospecies) and the

220 Bushfire and forest invertebrates

Table 3.11 Presence of Dipteran morphospecies on burnt and unburnt plots.

221 Australia’s Biodiveristy - Responses to Fire

Table 3.10 Breakdown of Dipteran morphospecies by sub-order and family. # Morphospecies Sub-order Family Common name unburnt burnt total Nematocera Sciaridae - 8 5 8 Cecidomyiidae gall midges 6 6 7 Ceratopogonidae sand flies 5 4 7 Chironomidae midges 3 2 3 Scatopsidae - 1 1 1 Tipulidae crane flies 1 0 1 Mycetophilidae fungus gnats 1 0 1 Brachycera Phoridae - 16 13 17 Empididae* - 6 3 7 Chloropidae - 5 4 6 Muscidae bush flies 4 3 5 Sphaeroceridae - 2 1 3 Tachinidae - 2 1 3 Dolichopodidae - 1 1 1 Drosophilidae vinegar flies 2 0 2 Therevidae - 1 0 1 Syrphidae hover flies 1 0 1 Micropezidae stilt-legged flies 1 0 1 Piophilidae - 0 1 1 Calliphoridae blowflies 0 1 1 Totals 66 46 77 * all morphospecies from sub-family Tachydromiinae sub-plot of a treatment, it is difficult to identify clear patterns from these tabulated data (at the species level). At this stage it is apparent however that unburnt plots have more species from the families Sciaridae (8 vs 5), Phoridae (16 vs 13) and Empididae (6 vs 3). Although this table of relative abundance enables broad “assemblages” of species with similar responses to disturbance to be identified, these data are more clearly displayed in the form of a “bi-plot” derived from the CCA ordination (see 2.5.3.3). This graphical display (Figure 3.22) shows the configuration of the environmental variables, the scatter of sub-plots, and the Figure 3.22 Bi-plot from CCA ordination of Dipteran relationship between the two, giving an overview presence/absence data. (Ellipses represent 1 standard of how community composition varies with the deviation unit around treatment centroids) environment (Ter Braak 1986). A number of features are apparent from this were on average 44% lower than on unburnt sub- bi-plot. Firstly, the substantial overlap of unburnt plots (see Section 3.3.3.2), however the slightly and burnt sub-plots in ordination space reflects higher similarity of unburnt sub-plots means that, the relatively large number of morphospecies (35) overall, burnt plots had 30% less species than shared by the two treatments. Secondly, the loose unburnt plots (46 vs 66). Unburnt sub-plots have clustering of both unburnt and burnt sub-plots higher richness (α-diversity) and are more similar, indicates similar within-treatment diversity. Burnt resulting in lower “turnover” between sub-plots sub-plots are slightly more diverse (spread-out) (lower β-diversity). indicating a higher turnover (β-diversity) for this The third feature concerns the contribution treatment. Richness values on burnt sub-plots of environmental variables to the differences in

222 Bushfire and forest invertebrates

species composition for the two treatments. In the Gnaphosidae and Corinnidae, containing 14, 13 bi-plot (Figure 3.22) the length of the arrow and 13% of overall morphospecies respectively signifies the relative contribution of that variable (see Table 3.12). Most (62%) of the families were to species composition, and the direction signifies represented by only 1 or 2 species. its contribution to the differences between Overall, burnt plots had a greater number of treatments. Unburnt sub-plots are characterised morphospecies than unburnt plots (48 and 31 by high levels of litter and high cover of tall and respectively), with 16 morphospecies (25%) very tall shrubs. Due to the correlation between common to both treatments (Group A - Table variables (see Section 3.2) these plots are also 3.13). When the morphospecies are arranged to characterised by high top-soil moisture levels and reflect their distribution across plots and low and spatially variable amounts of insolation at treatments, it is clear that there is a group of ground level. Burnt sub-plots are characterised by species (15) found only on unburnt sub-plots high levels of insolation at ground level, and to a (24% - Group B) and a different group of species lesser extent, steeper slopes and greater exposure (32) found only on burnt sub-plots (51% - Group (more north-westerly aspects). The amount of C). Frequent burning would therefore appear to litter and the level of insolation at ground level have resulted in the loss of up to 15 species of make the greatest contribution to the differences spider, however the changed environment supports in species composition between unburnt and up to 32 new species not found in unburnt areas. burnt plots. Other environmental variables make Morphospecies found on both treatments only a minor contribution to the observed were representative of the families Corinnidae differences in Dipteran species composition (3 species), Textricellidae and Zodariidae (2 species between burnt and unburnt treatments. each), Hahniidae, Linyphiidae, Gnaphosidae, 3.3.4.3 Spiders Ctenidae, Micropholcommatidae, Oonopidae, Thomisidae, Toxopidae and Theridiidae (1 species The 63 spider morphospecies were representative each). Morphospecies found only on unburnt sub- of 21 families, with the most diverse groups plots are representative of the families Theridiidae, overall being the families Zodariidae,

Table 3.12 Breakdown of spider morphospecies by family. # Morphospecies Family Common name Ecological Information unburnt burnt total Zodariidae - terrestrial; under stones, logs & litter 2 9 9 Gnaphosidae - terrestrial; under stones, logs & litter 3 6 8 Corinnidae - generalised hunters & ant specialists 4 7 8 Salticidae jumping spiders terrestrial hunters on foliage, trees & logs. 3 3 6 Linyphiidae tent spiders build webs in foliage & near ground 2 4 5 Theridiidae - web builders; near ground level 3 2 4 Textricellidae - moist habitats; litter dwellers 2 3 3 Clubionidae ant-mimicking spiders nocturnal hunters; live in rolled-up leaves & in litter 1 1 2 Tekellidae - moist habitats; litter dwellers 1 1 2 Oonopidae - cryptic; occur under stones, logs & litter. 2 1 2 Ctenidae - terrestrial; vagrant hunters on ground & in litter 1 1 1 Hahniidae - build sheet webs in litter & foliage 1 1 1 Micropholcommatidae - moist habitats; litter dwellers 1 1 1 Thomisidae crab/flower spiders occur on foliage & bark 1 1 1 Toxopidae - moist habitats 1 1 1 Lycosidae wolf spiders terrestrial; ground hunters 0 3 3 Prododomidae - dry habitats 0 1 1 Heteropodidae - bark, foliage & litter dwellers 0 1 1 Dictynoidea - sheet web builders on ground, litter & bark 0 1 1 Malkaridae - moist habitats; litter dwellers 2 0 2 Stiphidiidae - sheet web builders 1 0 1 TOTAL 31 48 63

223 Australia’s Biodiveristy - Responses to Fire Table 3.13 Presence of Spider morphospecies on burnt and unburnt plots. Table

224 Bushfire and forest invertebrates

Salticidae, Gnaphosidae and Malkaridae (2 species A number of features are apparent from this each), Oonopidae, Tekellidae, Stiphidiidae, bi-plot. Firstly, the minimal overlap of unburnt and Linyphiidae, Clubionidae and Corinnidae burnt sub-plots in ordination space reflects the (1 species each). Morphospecies found only on largely dissimilar species assemblages of these two burnt sub-plots are representative of the families treatments. Secondly, the tighter clustering of Zodariidae (7 species), Gnaphosidae (5 species), unburnt sub-plots indicates a lower within- Corinnidae (4 species), Salticidae and Lycosidae treatment diversity compared to burnt sub-plots. (3 species each), Linyphiidae (2 species) and Burnt sub-plots are slightly more diverse (spread- Tekellidae, Textricellidae, Theridiidae, out) indicating a higher turnover (β-diversity) for Clubionidae, Dictynoidea, Heteropodidae and this treatment. Richness values on burnt sub-plots Prododomidae (1 species each). were on average were 27% higher than on unburnt As many of these families are represented sub-plots (see Section 3.3.3.3), however the greater only by single individuals or found only on one dissimilarity of burnt sub-plots means that, overall, sub-plot of a treatment, it is difficult to identify burnt plots had 55% more species than unburnt clear patterns from these tabulated data (at the plots. Burnt sub-plots have higher richness (α- species level). At this stage it is apparent however diversity) and are less similar, resulting in higher that unburnt plots have more species from the “turnover” between sub-plots (higher β-diversity). family Malkaridae (2 vs 0), while burnt plots have The third feature concerns the contribution greater numbers of species from the families of environmental variables to the differences in Zodariidae (9 vs 2), Gnaphosidae (6 vs 3), species composition for the two treatments. In the Corinnidae (7 vs 4), Linyphiidae (4 vs 2) and bi-plot (Figure 3.23) the length of the arrow Lycosidae (3 vs 0). signifies the relative contribution of that variable to Although this table of relative abundance species composition, and the direction signifies its enables broad “assemblages” of species with contribution to the differences between treatments. similar responses to disturbance to be identified, Unburnt sub-plots are characterised by high levels these data are more clearly displayed in the form of litter and high cover of tall and very tall shrubs. of a “bi-plot” derived from the CCA ordination Due to the correlation between variables (see (see 2.5.3.3). This graphical display (Figure 3.23) Section 3.2) these plots are also characterised by shows the configuration of the environmental high top-soil moisture levels and low and spatially variables, the scatter of sub-plots, and the variable amounts of insolation at ground level. relationship between the two, giving an overview Burnt sub-plots are characterised by high levels of of how community composition varies with the insolation at ground level, and to a lesser extent, environment (Ter Braak 1986). steeper slopes and greater cover of the herb & shrub component of the understorey vegetation. The amounts of insolation and litter make the greatest contribution to the differences in species composition between unburnt and burnt plots. Other environmental variables make only a minor contribution to the observed differences in spider species composition. 3.3.4.4 Beetles The 139 beetle morphospecies were representative of nine super-families and 25 families (see Table 3.14). The most diverse groups were the families Staphylinidae and Curculionidae, containing 22% and 17% of overall morphospecies respectively. Nine super-families were represented: the Figure 3.23 Bi-plot from CCA ordination of spider Staphylinoidea (56 morphospecies), the presence/absence data. (Ellipses represent 1 standard Curculionoidea (23 morphospecies), Caraboidea, deviation unit around treatment centroids) Scarabaeoidea, Cucujoidea (12 morphospecies each), Chrysomeloidea (10 morphospecies), Tenebrionoidea (9 morphospecies), Bostrichoidea

225 Australia’s Biodiveristy - Responses to Fire

Table 3.14 Breakdown of Beetle morphospecies by super-family and family. # Morphospecies Super-family Family Common name unburnt burnt total Staphylinoidea Staphylinidae Rove beetles 21 20 31 Pselaphidae - 6 7 11 Scydmaenidae - 5 7 8 Ptiliidae - 3 1 3 Leiodidae - 3 1 3 Scarabaeoidea Scarabaeidae - 6 7 9 Trogidae - 1 0 1 Hybosoridae - 1 1 1 Passalidae - 0 1 1 Elateroidea Elateridae Click Beetles 1 1 2 Cucujoidea Nitidulidae - 2 1 3 Corylophidae - 1 2 3 Endomychidae - 1 1 2 Lathridiidae - 2 1 2 Silvanidae - 0 1 1 Phalacridae - 0 1 1 Chrysomeloidea Chrysomelidae Leaf Beetles 4 8 9 Cerambycidae Longicorn Beetles 0 1 1 Curculionoidea Curculionidae Weevils 9 18 23 Caraboidea Carabidae Ground Beetles 11 8 12 Bostrichoidea Anobiidae - 2 2 3 Tenebrionoidea Tenebrionidae Darkling Beetles 3 2 5 Aderidae - 2 0 2 Oedemeridae - 1 0 1 Anthicidae - 1 0 1 Totals 86 92 139

(3 morphospecies), and the Elateroidea on unburnt plots and nine on burnt plots. Many (2 morphospecies). Tenebrionoidea are scavengers inhabiting the litter Most Staphylinoidea are general predators, layer. Seven morphospecies were found on with some feeding on decomposing fruits. Thirty- unburnt plots and two on burnt plots. The eight morphospecies from this super-family were Bostrichoidea were represented by one family, the found on unburnt plots and thirty-six on burnt Anobiidae, which feed on a variety of plant & plots (see Table 3.14). The Curculionoidea were animal products. Three morphospecies from this represented by one family, Curculionidae (the family were each found, two on unburnt and two Weevils), with nine morphospecies found on on burnt plots. The Elateroidea were represented unburnt plots and eighteen on burnt plots. The by a single family, the Elateridae (Click Beetles), Caraboidea were represented by one family, which are mainly predatory. Two morphospecies Caraboidea (the Ground Beetles), which are mainly from this family were found, one on unburnt and predatory on plant-inhabiting insects. Eleven one on burnt plots (see Table 3.14). morphospecies from this family were found on Overall, slightly more morphospecies were unburnt plots and eight on burnt plots. The found on burnt plots compared with unburnt Scarabaeoidea commonly feed on eucalypt leaves as plots (92 and 86 respectively), although on average, adults. Eight morphospecies from this super-family burnt sub-plots had 27% fewer species (see Section were found on unburnt plots and nine on burnt 3.3.3.4). Thirty-nine morphospecies (28%) were plots. Many Cucujoidea are known to feed on fungi common to both treatments (Group A - Table growing on leaf surfaces. Six morphospecies from 3.15). When the morphospecies are arranged to this super-family were found on unburnt plots and reflect their distribution across plots and seven on burnt plots. The Chrysomeloidea feed on treatments, it is clear that there is a group of species leaves and other vegetative parts of plants, both as (47) found only on unburnt sub-plots (34% - larvae and adults. Four morphospecies were found Group B) and a different group of species (53)

226 Bushfire and forest invertebrates

found only on burnt sub-plots (38% - Group C). sub-plots suggests a lower within-treatment Frequent burning would therefore appear to have diversity compared to unburnt sub-plots (ie. a resulted in the loss of up to 47 species of Coleoptera, lower β-diversity). Richness values on burnt sub- however the changed environment supports up to 53 plots were, on average, 27% lower than on unburnt new species not found in unburnt areas. sub-plots (see Section 3.3.3.4), with the apparently Morphospecies found on both treatments are high similarity of sub-plots for this treatment representative of the families Staphylinidae meaning that the overall richness of burnt plots (10 species), Carabidae (7 species), Scarabaeidae, should be much lower than for unburnt plots. In Scydmaenidae, Curculionidae (4 species each), fact, overall, burnt plots had slightly more species Chrysomelidae (3 species), Pselaphidae (2 species), than unburnt plots (92 vs 86). This clustering is Ptiliidae, Hybosoridae, Lathridiidae, Anobiidae and therefore an artefact of the two-dimensional Leiodidae (1 species each). Morphospecies found representation of the data, with an inspection of only on unburnt sub-plots are representative of the Table 3.15 revealing a greater number of species families Staphylinidae (11 species), Curculionidae unique to burnt plots (53 vs 47). While unburnt (5 species), Carabidae, Pselaphidae (4 species each), sub-plots have higher richness (α-diversity), burnt Tenebrionidae (3 species), Aderidae, Scarabaeidae, plots are less similar, resulting in higher “turnover” Ptiliidae and Nitidulidae (2 species each), between sub-plots (higher β-diversity). Scydmaenidae, Chrysomelidae, Lathridiidae, The third feature concerns the contribution Anobiidae, Leiodidae, Trogidae, Elateridae, of environmental variables to the differences in Corylophidae, Endomychidae, Oedemeridae and species composition for the two treatments. In the Anthicidae (1 species each). Morphospecies found bi-plot (Figure 3.24) the length of the arrow only on burnt sub-plots are representative of the signifies the relative contribution of that variable to families Curculionidae (14 species), Staphylinidae species composition, and the direction signifies its (10 species) Chrysomelidae, Pselaphidae (5 species contribution to the differences between treatments. each), Scarabaeidae, Scydmaenidae (3 species each), Unburnt sub-plots are characterised by high levels Corylophidae, Tenebrionidae, (2 species each), of litter and high cover of tall and very tall shrubs. Passalidae, Elateridae, Nitidulidae, Endomychidae, Due to the correlation between variables (see Silvanidae, Phalacridae, Cerambycidae, Carabidae Section 3.2) these plots are also characterised by and Anobiidae (1 species each). high top-soil moisture levels and low and spatially As many of these families are represented variable amounts of insolation at ground level. only by single individuals or found only on one Burnt sub-plots are characterised by high levels of sub-plot of a treatment, it is difficult to identify insolation at ground level, and to a lesser extent, clear patterns from these tabulated data (at the steeper slopes. These environmental variables make species level). At this stage it is apparent however the greatest contribution to the differences in that unburnt plots have more species from the species composition between unburnt and burnt family Carabidae (11 vs 8), while burnt plots have plots, with other environmental variables make greater numbers of species from the families only a minor contribution to the observed Curculionidae (18 vs 9) and Chrysomelidae (8 vs 4). differences in beetle species composition. Although this table of relative abundance enables broad “assemblages” of species with similar responses to disturbance to be identified, these data are more clearly displayed in the form of a “bi-plot” derived from the CCA ordination (see 2.5.3.3). This graphical display (Figure 3.24) shows the configuration of the environmental variables, the scatter of sub-plots, and the relationship between the two, giving an overview of how community composition varies with the environment (Ter Braak 1986). A number of features are apparent from this bi-plot. Firstly, the small overlap of unburnt and burnt sub-plots in ordination space reflects the Figure 3.24 Bi-plot from CCA ordination of Beetle largely dissimilar species assemblages of these two presence/absence data. (Ellipses represent 1 standard treatments. Secondly, the tighter clustering of burnt deviation unit around treatment centroids)

227 Australia’s Biodiveristy - Responses to Fire Table 3.15. Presence of Beetle morphospecies on burnt and unburnt plots. Table

228 Bushfire and forest invertebrates

to be scanned

229 Australia’s Biodiveristy - Responses to Fire

3.3.4.5 Ants plots (70 and 68 respectively) with the sub-families The 88 ant morphospecies were representative of generally equally represented at the morphospecies 5 sub-families and 34 genera (see Table 3.16). The level: Myrmeciinae 3 & 4, Myrmicinae 18 & 20, most diverse groups overall were the genera Dolichoderinae 7 & 9, and Formicinae 26 & 22 Pheidole, Iridomyrmex and Camponotus, each respectively. The exception was the Ponerinae, with containing 8% of overall morphospecies. 22 morphospecies on unburnt plots and 11 on Five sub-families were represented: Ponerinae burnt plots (see Table 3.16). (11 genera, 19 morphospecies), Myrmeciinae (1 A substantial proportion (50) of genus, 5 morphospecies), Myrmicinae (11 genera, morphospecies (57%) were common to both 24 morphospecies), Dolichoderinae (4 genera, 10 treatments (Group A - Table 3.17). When the morphospecies) and Formicinae (7 genera, 30 morphospecies are arranged to reflect their morphospecies). Overall, a similar number of distribution across plots and treatments, it is clear morphospecies were found on unburnt and burnt that there is a group of species (20) found only on

Table 3.16 Breakdown of Ant morphospecies by sub-family and genus. # Morphospecies Sub-family Genus Ecological role* unburnt burnt total

Ponerinae Rhytidoponera opportunist 4 4 4 Bothroponera solitary forager 1 1 1 Trachymesopus cryptic in soil/litter 1 1 1 Cerapachys climate specialist 2 1 2 Hypoponera cryptic in soil/litter 4 1 4 Heteroponera cryptic in soil/litter 2 1 2 Amblyopone cryptic in soil/litter 1 0 1 Discothyrea cryptic in soil/litter 1 0 1 Ponera cryptic in soil/litter 1 0 1 Leptogenys specialist predator 0 1 1 Sphinctomyrmex cryptic in soil/litter 0 1 1 Myrmeciinae Myrmecia solitary forager 3 4 5 Myrmicinae Pheidole generalist 5 6 7 Solenopsis cryptic in soil/litter 3 2 3 Crematogaster generalist 2 2 2 Strumigenys cryptic in soil/litter 1 1 1 Mayriella opportunist 1 1 1 Meranoplus climate specialist 1 1 1 Tetramorium opportunist 2 1 2 Epopostruma specialist predator 0 1 1 Podomyrma climate specialist 1 2 2 Monomorium generalist 1 2 2 Colobostruma solitary forager 0 2 2 Dolichoderinae Iridomyrmex dominant 5 7 7 Tapinoma cryptic in soil/litter 1 1 1 Technomyrmex opportunist 1 1 1 Leptomyrmex dominant 0 1 1 Formicinae Camponotus sub-dominant 6 6 7 Paratrechina opportunist 4 3 4 Polyrhachis sub-dominant 5 3 6 Melophorus climate specialist 5 3 5 Prolasius climate specialist 3 2 3 Stigmacros cryptic in soil/litter 2 3 3 Notoncus climate specialist 1 2 2 Totals 70 68 88

* ecological functional groups as defined by Andersen (1990)

230 Bushfire and forest invertebrates

unburnt sub-plots (23% - Group B) and a different Morphospecies found on both treatments are group of species (18) found only on burnt sub-plots representative of the genera: Camponotus, (20% - Group C). Frequent burning would Iridomyrmex (5 species each), Rhytidoponera, Pheidole therefore appear to have resulted in the loss of up to (4 species each), Paratrechina, Melophorus (3 species 20 species of ants, however the changed each), Myrmecia, Solenopsis, Crematogaster, Prolasius, environment supports up to 18 new species not Polyrhachis, Stigmacros (2 species each), Bothroponera, found in unburnt areas. Trachymesopus, Cerapachys, Hypoponera, Heteroponera, Table 3.17. Presence of Ant morphospecies on burnt and unburnt plots.

231 Australia’s Biodiveristy - Responses to Fire

Strumigenys, Mayriella, Meranoplus, Tetramorium, Podomyrma, Monomorium, Tapinoma, Technomyrmex and Notoncus (1 species each). Morphospecies found only on unburnt sub-plots are representative of the genera: Polyrhachis, Hypoponera (3 species each), Melophorus (2 species), Cerapachys, Heteroponera, Amblyopone, Discothyrea, Ponera, Myrmecia, Pheidole, Solenopsis, Tetramorium, Camponotus, Paratrechina and Prolasius (1 species each). Morphospecies found only on burnt sub-plots are representative of the genera: Iridomyrmex, Myrmecia, Pheidole, Colobostruma (2 species each), Rhytidoponera, Leptogenys, Sphinctomyrmex, Epopostruma, Figure 3.25 Bi-plot from CCA ordination of Ant relative Podomyrma, Monomorium, Leptomyrmex, Camponotus, abundance data. (Ellipses represent 1 standard Polyrhachis, Stigmacros and Notoncus (1 species each). deviation unit around treatment centroids) As many of these families are represented only by single individuals or found only on one indicates a lower within-treatment diversity sub-plot of a treatment, it is difficult to identify compared to unburnt sub-plots (ie. a lower β- clear patterns from these tabulated data (at the diversity). Richness values on burnt sub-plots were individual species level). At this stage it is apparent on average 26% higher than on unburnt sub-plots however that unburnt plots have more species (see Section 3.3.3.5), however the high similarity of from the genera Cerapachys (7 vs 1) and Hypoponera sub-plots for this treatment means that the overall (4 vs 1), while burnt plots have greater numbers richness of unburnt and burnt plots was similar (70 of species from the genus Colobostruma (2 vs 0). vs 68 species). Unburnt sub-plots have lower Although this table of relative abundance richness (α-diversity) but are less similar, resulting enables broad “assemblages” of species with in higher “turnover” between sub-plots (higher β- similar responses to disturbance to be identified, diversity). these data are more clearly displayed in the form The third feature concerns the contribution of a “bi-plot” derived from the CCA ordination of environmental variables to the differences in (see 2.5.3.3). This graphical display (Figure 3.25) species composition for the two treatments. In the shows the configuration of the environmental bi-plot (Figure 3.25) the length of the arrow variables, the scatter of sub-plots, and the signifies the relative contribution of that variable to relationship between the two, giving an overview species composition, and the direction signifies its of how community composition varies with the contribution to the differences between treatments. environment (Ter Braak 1986). Unburnt sub-plots are characterised by high A number of features are apparent from this levels of litter and high cover of tall and very tall bi-plot. Firstly, the lack of any overlap of unburnt shrubs. Due to the correlation between variables and burnt sub-plots in ordination space reflects the (see Section 3.2) these plots are also characterised largely dissimilar species assemblages of these two by high top-soil moisture levels and low and treatments. Although 57% of morphospecies were spatially variable amounts of insolation at ground found on both treatments, their relative abundance level. Burnt sub-plots are characterised by high on each treatment differs substantially, leading to levels of insolation at ground level, and to a lesser fundamentally different species assemblages. extent, steeper slopes, greater cover of the herb & Secondly, the tighter clustering of burnt sub-plots shrub component of the understorey vegetation,

Table 3.18 Distribution of invertebrate morphospecies richness by treatment Invertebrate both unburnt only burnt only total Taxa # % # % # % # Hemiptera (bugs) 7 16 18 41 19 43 44 Diptera (flies) 35 45 31 40 11 15 77 Araneae (spiders) 16 25 15 24 32 51 63 Coleoptera (beetles) 39 28 47 34 53 38 139 Formicidae (ants) 50 57 20 23 18 20 88 Total 147 - 131 - 133 - 411

232 Bushfire and forest invertebrates

greater top-soil hardness and greater spatial biodiversity of frequently burnt areas was patchiness of twigs (0–10mm) and very tall shrubs. maintained by the addition of species not The level of insolation at ground level, the recorded on unburnt plots. The changed amount of litter and the cover of tall shrubs make the environment was supporting an additional 133 greatest contribution to the differences in species morphospecies (19 bugs, 11 flies, 32 spiders, 53 composition between unburnt and burnt plots. beetles and 18 ants). Other environmental variables make only a minor Many of the morphospecies apparently lost contribution to the observed differences in ant species. from frequently burnt sites were however only 3.3.4.6. Summary detected on a single sub-plot or represented by a single individual on unburnt plots. These could be The five groups studies in detail proved to be genuinely rare or uncommon species which were extremely diverse. Beetles had the richness fauna missed purely by chance when sampling burnt overall with 139 beetle morphospecies plots. For this reason it is difficult to identify clear representative of nine super-families and 25 patterns (at the species level) from the relative families. The ants were the second richness group abundance data alone. Some general trends were with 88 morphospecies representative of 5 sub- apparent however when morphospecies data were families and 34 genera. They were followed by the grouped into a higher taxonomic level. For bugs, flies with 77 morphospecies representative of 2 unburnt plots had more species from the infra- sub-orders and 20 families, the spiders with 63 order Dipsocoromorpha (7 vs 3), while burnt plots morphospecies representative of 21 families, and have greater numbers of species from the family the bugs with 44 morphospecies representative of Reduviidae (5 vs 0). For flies, unburnt plots had 16 family (or similar) groups. more species from the families Sciaridae (8 vs 5), Overall, the same number of morphospecies Phoridae (16 vs 13) and Empididae (6 vs 3). For (279) were collected from unburnt and burnt spiders, unburnt plots had more species from the plots. Richness on unburnt sub-plots was, on family Malkaridae (2 vs 0), while burnt plots had average, 48.2 morphospecies, which was similar to greater numbers of species from the families the average value on burnt sub-plots (46.5 Zodariidae (9 vs 2), Gnaphosidae (6 vs 3), morphospecies). This initially suggests that Corinnidae (7 vs 4), Linyphiidae (4 vs 2) and frequent burning had not reduced biodiversity in Lycosidae (3 vs 0). For beetles, unburnt plots had this forest environment. An analysis of the more species from the family Carabidae (11 vs 8), richness of individual groups (see Section 3.3.3) while burnt plots had greater numbers of species has shown this not to be the case, with groups from the families Curculionidae (18 vs 9) and responding differently to frequent burning. This Chrysomelidae (8 vs 4). For ants, unburnt plots Section (3.3.4) examined the nature of that had more species from the genera Cerapachys (7 vs response by looking at the composition of faunal 1) and Hypoponera (4 vs 1), while burnt plots had assemblages (communities). greater numbers of species from the genus An examination of the distribution of Colobostruma (2 vs 0). The implication of these morphospecies across sub-plots for each changes for community organisation and treatment, detected a consistent pattern. ecosystem function are considered in Section 3.4. Morphospecies fell into one of three groups: found Although an examination of relative abundance on both treatments (Group A), found only on patterns enables broad “assemblages” of species with unburnt plots (Group B), or found only on burnt similar responses to disturbance to be identified, plots (Group C). The relative proportions of these data are more clearly displayed and interpreted morphospecies in each category however varied in the form of a “bi-plot” derived from the CCA substantially between taxonomic groups. For ordination. These graphical displays show the Hemiptera (bugs) the proportions were 16, 41 and configuration of the environmental variables, the 43% for both, unburnt and burnt respectively; for scatter of sub-plots, and the relationship between the Diptera (flies) 45, 40 and 15%; for spiders 25, 24 two, giving an overview of how community and 51%; for beetles 28, 34 and 38%; and for ants composition varies with the environment. An 57, 23 and 20% (see Table 3.18). examination of the bi-plots revealed a number of These results also suggest that frequent consistent features. Firstly, the degree of overlap of burning had led to the loss of up to 131 species unburnt and burnt sub-plots in ordination space (18 bugs, 31 flies, 15 spiders, 47 beetles and 20 reflected the similarity (or dissimilarity) of the ants), which represents 47% of the morphospecies species’ assemblages of the two treatments. For bugs, known from the unburnt areas. Overall 233 Australia’s Biodiveristy - Responses to Fire

spiders, beetles and ants there was little or no level and, to a lesser extent, steeper slopes. These overlap, indicating low similarity of the two environmental variables make the greatest assemblages. For flies however there was a contribution to the differences in species composition substantial overlap, reflecting the relatively large between unburnt and burnt plots. For bugs, burnt number of morphospecies shared by the two sub-plots were also characterised by greater top-soil treatments. hardness and greater cover of the herb & shrub Secondly, the degree of clustering of the sub- component of the understorey vegetation, with herb plots from each treatment indicates the relative and shrub cover also important for spiders and ants. similarity of species’ assemblages on sub-plots and For flies, burnt sites were also characterised by plots within each treatment. The tighter clustering greater exposure (more north-westerly aspects). of burnt sub-plots for bugs, beetles and ants Other environmental variables made only a minor indicated a lower within-treatment diversity contribution to the observed differences in species compared to unburnt sub-plots (ie. a lower β- composition between burnt and unburnt treatments diversity). The converse applied for spiders, with for all taxa. Communities were therefore influenced the tighter clustering of unburnt sub-plots by a combination of site-dependent (slope and aspect) indicating a lower β-diversity compared to burnt and treatment-dependent (litter, insolation, herb & sub-plots. The situation for flies indicated similar shrub cover, top-soil moisture & hardness) within-treatment diversity for both treatments, with environmental variables. loose clustering of both unburnt and burnt sub- plots. This interaction between point richness 3.3.5 Community Structure (α-diversity) and spatial “turnover” of species While it is possible to describe and assess (β-diversity) has substantial implications for the communities using indices such as species interpretation of the apparent effect of repeated richness, or to compare the relative abundance of burning and were previously identified as “scale species using similarity indices, multi-variate effects” in Section 3.3.3.6. Similar patterns were approaches and/or through graphical apparent for bugs and ants where richness values representation, these contribute little to an (α-diversity) on burnt sub-plots were on average understanding of the processes underlying their higher than on unburnt sub-plots, however the differences. In order to simplify and interpret the high similarity of assemblages on burnt sub-plots complexity of ecological systems, one approach (low β-diversity) meant that the overall richness of has been to group species into “guilds” or both treatments were similar. Unburnt sub-plots “functional groups”. These groups recognise the had lower richness (α-diversity) but are less similar, ecological rather than the taxonomic affinity of resulting in higher “turnover” between sub-plots species. In this Section morphospecies were (higher β-diversity), increasing overall species allocated to guilds following reference to the richness for that treatment. For flies and beetles relevant literature and discussions with taxonomic burnt plots had lower richness (α-diversity) but experts. To maintain comparability between broad were less similar, resulting in higher β-diversity. taxonomic groups, the number of species from Spiders exhibited a different pattern with both each group was averaged across the 24 sub-plots higher α- and β-diversity for the burnt treatment, for each treatment. This method does not take resulting in a large number of species (32) unique to into account the relative abundance of individuals frequently burnt sites. of species and therefore is not unfairly biased by a The third feature of the bi-plots concerns the few very abundant species. contribution of environmental variables to the 3.3.5.1 Hemiptera differences in species composition for the two Morphospecies were classified into one of 5 groups treatments. In the bi-plots (Figures 3.21-5) the length based on the known habits and requirements at the of the arrows signify the relative contribution of that family level (see Table 3.19). These groups were variable to species composition, and the direction primarily based on feeding strategy, and secondarily signifies their contribution to the differences between habitat preferences. The groups were; primarily treatments. For all taxa unburnt sub-plots were phytophagous (Cicadellidae, Coccidae, Fulgoridae characterised by high levels of litter, high cover of tall and the unidentified Homopteran), moist habitat and very tall shrubs, high top-soil moisture levels and specialists (Dipsocoromorpha), primarily predacious low and spatially variable amounts of insolation at (Nabidae, Reduvidae) and others. This last category ground level. Similarly, burnt sub-plots were included morphospecies from the families characterised by high levels of insolation at ground Enicocephalidae, Anthocoridae, Miridae, Tingidae, 234 Bushfire and forest invertebrates

Table 3.19 Comparison of Hemipteran community structure on burnt and unburnt plots. Data represent mean percentage of morphospecies in each category. Ecological role Taxa Unburnt Burnt Primarily phytophagous Cicadellidae 48.3 75.7 Coccidae, Fulgoridae, & Homoptera 15.4 3.3 Moist habitat specialists Dipsocoromorpha 22.9 3.9 Primarily predacious Nabidae 11.7 0.0 Reduvidae 0.0 10.1 Others (see text) 1.7 7.0 Total 100 100

Thaumastocoridae, Lygaeidae, Plataspididae and family Reduviidae. The greatest change concerns Pentatomidae which were only found on one sub- habitat specialists from the sub-order plot and therefore offering limited information to Dipsocoromorpha where there has been, on this analysis. average, an 83% reduction in the number of On average, 64% of morphospecies on morphospecies. unburnt plots were primarily phytophagous (plant 3.3.5.2 Diptera eating) and 12% predacious on other invertebrates. Morphospecies were classified into one of 7 groups Of the remaining 24%, approximately 23% are based on the known habits and requirements at the known to be moist habitat specialists from the sub- family level (see Table 3.20). These groups were order Dipsocoromorpha utilising a variety of primarily based on feeding strategy, and feeding strategies. On average, 79% of secondarily habitat preferences. The groups were; morphospecies on burnt plots were primarily primarily phytophagous (Cecidomyiidae), phytophagous and 10% predacious on other primarily predacious (Tachydromiinae), fungal invertebrates. Of the remaining 11%, approximately feeders (Sciaridae, Drosophilidae, Mycetophilidae 4% are known to be moist habitat specialists from & Scatopsidae), generalists and scavengers the sub-order Dipsocoromorpha. The remaining (Phoridae & Chloropidae), moist habitat specialists 7% of morphospecies on burnt plots were from the (Ceratopogonidae & Chironomidae), litter families Anthocoridae, Miridae, Thaumastocoridae, dwellers (Sphaeicieiidae & Tipilidae) and wide- Lygaeidae, Plataspidae and Pentatomidae (1 species ranging “tourists” (Piophilidae, Micropezidae, each). Dolichopodidae, Calliphoridae, Syrphidae, A graphical comparison of these data (Figure Thereuidae, Muscidae & Tachinidae). 3.26) indicates that frequent burning has resulted On average, 29% of morphospecies on in a fundamental shift in Hemipteran community unburnt plots were primarily generalists and structure. With regard to feeding strategy, there scavengers, 23% feeders on fungal products, 17% has been, on average, a 15% increase in the moist habitat specialists, and 11% phytophagous number of phytophagous species. While the (plant eating). Of the remaining 20%, number of predacious species has remained approximately 13% are known to be wide-ranging largely unchanged, there has been a total shift “tourists”, 6% predators, and 2% litter dwellers. from species from the family Nabidae to the On average, 26% of morphospecies on unburnt plots were primarily generalists and scavengers, 13% feeders on fungal products, 18% moist habitat specialists, and 26% phytophagous (plant eating). Of the remaining morphospecies, approximately 7% are known to be wide-ranging “tourists”, 6% predators, and 1% litter dwellers. A graphical comparison of these data (Figure 3.27) indicates that frequent burning has resulted in a fundamental shift in Dipteran community structure. With regard to feeding strategy, there has been, on average, a 44% decrease in the number of morphospecies feeding on fungal Figure 3.26 Comparison of Hemipteran community products. This change was most marked in the structure on burnt and unburnt plots families Scaridae and Scatopsidae. The proportion

235 Australia’s Biodiveristy - Responses to Fire

Table 3.20 Comparison of Fly community structure on burnt and unburnt plots. Data represent mean percentage of morphospecies in each category. Ecological role Taxa Unburnt Burnt Fungal feeders Sciaridae, Drosophilidae, Mycetophilidae & Scatopsidae 23.2 12.9 Litter dwellers Sphaeicieiidae & Tipilidae 2.0 0.8 Tourists Piophilidae, Micropezidae, Dolichopodidae, Calliphoridae, Syrphidae, Thereuidae, Muscidae & Tachinidae 12.6 7.2 Moist habitat specialists Ceratopogonidae & Chironomidae 17.0 17.8 Generalists/scavengers Phoridae & Chloropidae 29.1 25.9 Predators Tachydromiinae 5.8 5.8 Phytophagous Cecidomyiidae 10.3 25.8 Total 100 100

With regard to micro-habitat preferences, the number of morphospecies specifically utilising the litter layer has, on average, decreased by 60%. This was primarily due to the absence of the family Tipulidae on frequently burnt sub-plots. The number of moist habitat specialists remained similar, although the family Ceratopogonidae was more commonly represented on burnt sub-plots and the family Chironomidae on unburnt sub-plots. 3.3.5.3 Spiders Morphospecies were classified into one of 5 groups Figure 3.27 Comparison of Fly community structure on based on the known habits and requirements at the burnt and unburnt plots family level (see Table 3.21). As all spiders are fundamentally predacious, these groups were of predacious species on sub-plots has remained primarily based on habitat preferences. The groups similar, however the number of phytophagous were; moist habitat specialists (Theridiidae, species has, on average, increased by 140%. These Toxopidae, Oonopidae, Malkaridae & Tekellidae), additional species were from the family those with a known preference for dry habitats Cecidomyiidae. The number of generalists and (Gnaphosidae & Corinnidae), litter dwellers scavengers has decreased slightly (11%). The (Hahniidae & Textricellidae), open & disturbed number of morphospecies regarded as wide- habitat specialists (Linyphiidae & Zodariidae) and ranging “tourists” was, on average, reduced by others. This last category included generalised 43% on frequently burnt plots. hunters from the families Ctenidae, Dictynoidea, Heteropodidae, Micropholcommatidae,

Table 3.21 Comparison of Spider community structure on burnt and unburnt plots. Data represent mean percentage of morphospecies in each category. Ecological role Family Unburnt Burnt Theridiidae 22.2 1.8 Moist habitat Toxopidae 11.4 1.2 specialists Oonopidae 8.1 0.4 Malkaridae 4.0 0.0 Tekellidae 4.3 2.2 Dry habitat Gnaphosidae 5.8 8.1 preference Corinnidae 14.6 18.4 Litter Hahniidae 4.5 10.4 dwellers Textricellidae 5.5 10.3 Open/disturbed Linyphiidae 3.1 15.7 habitat specialists Zodariidae 2.1 20.1 Others (see text) 14.3 11.3 Total 100 100

236 Bushfire and forest invertebrates

Stiphidiidae, Clubionidae, Lycosidae, Prododomidae, Salticidae and Thomisidae. Many of these were only found on one sub-plot, therefore offering limited information to this analysis. On average, 50% of morphospecies on unburnt plots were primarily moist habitat specialists, 20% have a known preference for dry habitats, 10% are litter dwellers, and 5% are known open/disturbed habitat specialists. The remaining 14% are generalist hunters with more flexible habitat requirements. On average, 6% of morphospecies on unburnt plots were primarily moist habitat specialists, 27% have a known preference for dry habitats, 21% are litter dwellers, Figure 3.28 Comparison of Spider community structure and 36% are known open/disturbed habitat on burnt and unburnt plots specialists. The remaining 11% are generalist primarily based on feeding strategy and were; hunters with more flexible habitat requirements. primarily predacious (Staphylinidae, Scydmaenidae, A graphical comparison of these data (Figure Pselaphidae and Carabidae), fungal feeders 3.28) indicates that frequent burning has resulted in (Leiodidae), generalists (Scarabaeidae), a fundamental shift in spider community structure. phytophagous (Chrysomelidae & Curculionidae) With regard to habitat preference, there has been, and others. This last category included on average, an 88% decrease in the number of morphospecies from the families Ptiliidae, moist habitat specialists, and an 35% increase in the Silvanidae, Endomychidae, Corylophidae, number of species with a known preference for dry Phalacridae & Lathridiidae (fungal feeders), habitats. The change in moist habitat specialists Trogidae, Hybosoridae, Elateridae, Nitidulidae, was primarily due to a 90-95% decrease in species Tenebrionidae and Anthicidae (generalists), and from the families Theridiidae, Toxopidae and Cerambycidae, Oedemeridae, Aderidae & Oonopidae. The number of litter dwelling species Anobiidae (phytophagous). They were not included has, on average, increased by 110% with an directly into the above groupings because most equivalent increase from the families Hahniidae morphospecies were only found on one sub-plot and Textricellidae. The number of species known and therefore offer limited information to this to prefer open and disturbed habitats has, on analysis. average, increased by over 600% due primarily to 7 On average, 70% of morphospecies on species from the family Zodariidae which were only unburnt plots were primarily predacious, 5% found on burnt sub-plots. fungal feeders, 7% generalists and 5% 3.3.5.4 Beetles phytophagous (plant eating). On average, 57% of Morphospecies were classified into one of 5 groups morphospecies on unburnt plots were primarily based on the known habits and requirements at the predacious, 3% fungal feeders, 12% generalists family level (see Table 3.22). These groups were and 16% phytophagous.

Table 3.22 Comparison of Beetle community structure on burnt and unburnt plots. Data represent mean percentage of morphospecies in each category. Ecological role Family Unburnt Burnt Staphylinidae 33.1 15.3 General Scydmaenidae 5.6 6.3 predators Pselaphidae 4.0 5.6 Carabidae 27.0 29.7 Fungal feeders Leionidae 4.8 3.3 Generalists Scarabaeidae 6.9 12.4 Phytophagous Chrysomelidae 1.3 6.6 Curculionidae 3.3 9.5 Others (see text) 14.0 11.3 Total 100 100

237 Australia’s Biodiveristy - Responses to Fire

A graphical comparison of these data (Figure Table 3.23 Comparison of Ant community 3.29) indicates that frequent burning has resulted in structure on burnt and unburnt plots. Data a fundamental shift in beetle community structure. represent mean percentage of morphospecies in With regard to feeding strategy, there has been, on each category. average, an 18% decrease in the number of predator Ecological role Taxa Unburnt Burnt species, due primarily to a 54% reduction in the Dominants Iridomyrmex 15.6 12.0 number of morphospecies from the family Leptomyrmex 0.0 0.7 Staphylinidae. The average number of Sub-dominants Camponotus 8.7 10.7 morphospecies feeding on fungal products (as Polyrhachis 1.3 1.8 represented by the family Leiodidae) has declined Cerapachys 1.2 0.8 by 31%, with some other fungal feeders (families Melophorus 3.3 3.3 Ptiliidae & Endomychidae) decreasing by over Climate Meranoplus 2.1 3.1 200%. The generalists, as represented by the family Specialists Notoncus 2.1 4.5 Podomyrma 0.2 0.4 Scarabaeidae, have increased on average by 79%. Prolasius 8.2 6.2 Phytophagous species have increased dramatically Amblyopone 0.5 0.0 (up by 250%), particularly morphospecies from the Discothyrea 0.2 0.0 family Chrysomelidae. For this family, 56% of its Heteroponera 1.6 1.1 overall morphospecies were only found on Hypoponera 1.7 0.2 frequently burnt sub-plots. Cryptic Ponera 0.2 0.0 Species Solenopsis 4.6 4.4 Sphinctomyrmex 0.0 0.4 Stigmacros 3.6 4.1 Tapinoma 2.5 0.4 Trachymesopus 0.8 2.3 Mayriella 0.9 0.8 Paratrechina 9.9 8.4 Opportunists Rhytidoponera 3.7 10.5 Technomyrmex 0.5 0.2 Tetramorium 3.8 4.6 Crematogaster 4.9 3.9 Generalists Monomorium 0.8 3.8 Pheidole 11.5 7.4 Bothroponera 2.8 0.7 Colobostruma 0.0 0.6 Solitary/ Epopostruma 0.0 0.2 Figure 3.29 Comparison of Beetle community structure Specialists Leptogenys 0.0 0.8 on burnt and unburnt plots Myrmecia 1.4 2.5 Total 100 100 3.3.5.5 Ants Morphospecies were classified into one of 7 On average, 16% of morphospecies on “functional groups” groups based on the known unburnt plots were dominants, 10% sub- habits and requirements at the generic level (see dominants, 18% climate specialists, 16% cryptic Andersen 1990). These groups were; dominants species, 19% opportunists, 17% generalists, and 4% (Iridomyrmex & Leptomyrmex), sub-dominants solitary/ specialist species. On average, 16% of (Camponotus & Polyrhachis), climate specialists morphospecies on unburnt plots were dominants, (Cerapachys, Melophorus, Meranoplus, Notoncus, 10% sub-dominants, 18% climate specialists, 16% Podomyrma & Prolasius), cryptic species cryptic species, 19% opportunists, 17% generalists, (Amblyopone, Discothyrea, Heteroponera, Hypoponera, and 4% solitary/specialist species. Ponera, Solenopsis, Sphinctomyrmex, Stigmacros, A graphical comparison of these data (Figure Tapinoma and Trachymesopus), opportunists 3.30) indicates that frequent burning has resulted in (Mayriella, Paratrechina, Rhytidoponera, a slight shift in ant community structure. The Technomyrmex, Tetramorium, Crematogaster, number of morphospecies within the “dominant” Monomorium & Pheidole) and solitary/specialist functional group has, on average, decreased by 23%, species (Bothroponera, Colobostruma, Epopostruma, primarily through a reduction in the occurrence of Leptogenys & Myrmecia). The average number of Iridomyrmex species. The number of “sub- morphospecies from these genera on each dominants” has increased slightly (25%), largely treatment are shown in Table 3.23. because of an increase in the occurrence of 238 Bushfire and forest invertebrates

3.3.5.6 Summary This Section examined community structure by grouping morphospecies into ecological groups based upon their feeding strategies and habitat preferences. By classifying large numbers of species into smaller, more manageable groups, it is possible to substantially reduce the apparent complexity of ecological systems and provide a basis for evaluating environmental change (Andersen 1990). For all groups studied, frequent burning Figure 3.30 Comparison of Ant community structure on resulted in a change in the structure of the burnt and unburnt plots community. With regard to feeding strategy, there was on average, a 15, 140 and 250% increase Camponotus species. With regard to “climate (respectively) in the number of phytophagous specialists”, they have remained largely stable (plant feeding) species of bugs, flies and beetles. (increased by 4%) as a group, although individual With regard to the proportion of predator species, genera did vary in their response (see Table 3.23). it remained unchanged for flies and bugs, however There was a decrease (19%) in the average number for bugs there was a total shift from the family of “cryptic” morphospecies, particularly with regard Nabidae to the family Reduviidae. With beetles to the genera Hypoponera (90%) and Tapinoma there was, on average, an 18% decrease in the (86%). A number of morphospecies from this group number of predator species, due primarily to a 54% however increased in their average occurrence; reduction in the number of morphospecies from Stigmacros (13%) and Trachymesopus (196%), the family Staphylinidae. For groups feeding although the actual number of species involved is primarily on fungal products there was, on average, small (overall 3 and 1 respectively). a 44% (flies) and 31% (beetles) decrease in the There was a substantial increase overall in number of morphospecies. For flies this was most the average number of “opportunist” marked in the families Scaridae and Scatopsidae, morphospecies (30%), primarily due to one genus, and for beetles, in the families Leiodidae, Ptiliidae Rhytidoponera, which increased by 180%. In fact & Endomychidae. For flies, the number of this was attributable to a single species: generalists and scavengers has decreased slightly Rhytidoponera metallica, which was extremely (11%), while for beetles the proportion of numerous on burnt sub-plots and contributed to generalists has increased, on average, by 79%, the high ant abundance detected for this treatment primarily through an increase in morphospecies (see Section 3.3.2.10). The number of “generalist” from the family Scarabaeidae. With ants, the morphospecies, on average, decreased by 11% proportion of generalist morphospecies decreased, following frequent burning. Results were not on average, by 11% following frequent burning. consistent within the group with the genera Results were not consistent within the group with Pheidole and Crematogaster decreasing by 35 and the genera Pheidole and Crematogaster decreasing by 19% respectively, and the genus Monomorium 35 and 19% respectively, and the genus increasing by 388%. The number of larger Monomorium increasing by 388%. For flies, the “solitary/specialist” morphospecies increased number of morphospecies regarded as wide- slightly (13%), although the results were quite ranging “tourists” was, on average, reduced by 43% variable within the group. Three new genera were on frequently burnt plots. found on burnt plots (Colobostruma, Epopostruma & When groups were compared with regard to Leptogenys) while the numbers of species of habitat preferences, it was apparent that there Myrmecia increased, on average, by 87%. The were substantial changes in community structure occurrence of the solitary forager Bothroponera sp. for some taxa. For bugs, numbers of moist habitat A decreased substantially (76%). specialists from the sub-order Dipsocoromorpha were, on average, reduced by 83%. Amongst spiders, moist habitat specialists were reduced by 88%, primarily due to a 90-95% decrease in numbers of morphospecies from the families

239 Australia’s Biodiveristy - Responses to Fire

Theridiidae, Toxopidae and Oonopidae. With the morphospecies, due primarily to a single species: flies, the number of moist habitat specialists Rhytidoponera metallica, which was extremely remained similar, although the family numerous on burnt sub-plots. Ceratopogonidae was more commonly It has been observed that the structure of ant represented on burnt sub-plots and the family communities, in particular, may be influenced by Chironomidae on unburnt sub-plots. The the relative abundance of particular “dominant” proportion of temperature-dependent (climate) and “sub-dominant’ groups (Fox & Fox 1982, specialists among the ants remained largely stable Andersen 1990). Following frequent fire, the (increased by 4%), although individual genera did number of morphospecies within the “dominant” vary in their response. For spiders, there was, on functional group had, on average, decreased by average, and an 35% increase in the number of 23%, primarily through a reduction in the species with a known preference for dry habitats. occurrence of Iridomyrmex species. The number of For groups primarily inhabiting the litter layer “sub-dominants” had increased slightly (25%), there was a variety of responses to frequent burning. largely because of an increase in the occurrence of For flies, the number of morphospecies specifically Camponotus species. The number of larger utilising the litter layer had, on average, decreased “solitary/specialist” morphospecies, which interact by 60%. This was primarily due to the absence of only slightly with other groups, had increased the family Tipulidae on frequently burnt sub-plots. slightly (13%), although the results were quite With ants there was a decrease (19%) in the average variable within the group. Three new genera were number of “cryptic” morphospecies inhabiting the found on burnt plots (Colobostruma, Epopostruma litter and soil, particularly with regard to the genera & Leptogenys) while the numbers of species of Hypoponera (90%) and Tapinoma (86%). For spiders Myrmecia increased, on average, by 87%. The however, the number of litter dwelling species occurrence of a solitary forager, Bothroponera sp.A, increased, on average, by 110% with an equivalent had decreased substantially (76%). increase from the families Hahniidae and Textricellidae. 3.3.6 Biodiversity Indicators Many groups of terrestrial invertebrates are It is often postulated that one group of invertebrates adapted to exploit disturbed habitats. For spider may act as an “indicator” or “umbrella” group for morphospecies known to prefer open and others, thereby allowing inferences to be made on disturbed habitats, the number of morphospecies the impact of disturbance regimes. The relationship increased, on average, by over 600% following between species richness of selected taxa was frequent burning, This was due primarily to the investigated here using Pearson’s Product-Moment occurrence of seven species from the family correlation analyses. Table 3.24 gives the correlation Zodariidae only on burnt sub-plots. For the ants coefficients (and probability values) for the there was a substantial increase overall (30%) in relationship between richness values for pairs of taxa the average proportion of “opportunist” at the sub-plot scale. The values in the bottom left

Table 3.24 Correlations between species richness values for five taxa. Values on the bottom-left represent standard Pearson’s Product-Moment correlation coefficients (n=48), those on the top-right are partial coefficients (n=43) controlling for the effects of the other taxa in each comparison. Pairs of data represent correlation coefficient (top) and probability (bottom) values. UNBURNT ANTS BEETLES BUGS SPIDERS FLIES ANTS -0.053 0.131 0.292 -0.343 0.727 0.391 0.052 0.021 BEETLES -0.238 0.071 -0.160 0.352 0.103 0.643 0.293 0.018 BUGS 0.174 0.005 0.055 -0.058 0.2360.974 0.722 0.704 SPIDERS 0.312 -0.189 0.093 0.122 0.031 0.199 0.529 0.424 FLIES -0.394 0.395 -0.098 -0.062 0.0060.0060.5060.676

240 Bushfire and forest invertebrates

of the table represent standard correlation coefficients, while those in the top-right are “partial” coefficients, controlling for the possible effect of other variables (taxa). This tests, for example, whether a spurious relationship between two taxa exists because of a separate relationship they may have individually with another taxa. A examination of the correlation coefficients reveals three significant relationships. Firstly, species richness values for spiders and ants are positively correlated (r = 0.312, P = 0.031, n = 48). This implies that as the richness of ants increases so does the richness of spiders. One could be used to predict the other, however the predictive power Figure 3.31 Relationship between species richness is low with only 9.6% of the variance in one taxa values for ants and spiders contributed by the variance of the other. In addition, this coefficient decreases slightly when controlling for the effect of other taxa (r = 0.292, ants, and between flies and beetles. Given this P = 0.052, n = 43). situation, and the previously identified differences Secondly, species richness values for flies and between burnt and unburnt areas, both in regard ants are negatively correlated (r = -0.394, P = to habitat conditions (see Section 3.1) and their 0.006, n = 48). This implies that as the richness of faunal assemblages (see Section 3.3), it was ants increases the richness of flies decreases (and appropriate to test correlations separately for the vice versa). One could be used to predict the two treatments. other, however the predictive power is low with Table 3.25 gives the correlation coefficients only 15.5% of the variance in one taxa (and probability values) for the relationship between contributed by the variance of the other. In richness values for pairs of taxa. The values in the addition, this coefficient decreases slightly when bottom-left of the table represent correlation controlling for the effect of other taxa (r = -0.343, coefficients for burnt sites, while those in the top- P = 0.021, n = 43). right are coefficients for unburnt sites. In other Thirdly, species richness values for flies and words, this approach “controls for” the possible beetles are positively correlated (r= 0.395, P = 0.006, effect of treatment. n = 48). This implies that as the richness of flies When the data are analysed in this way it is increases so does the richness of beetles. One could obvious that there are no significant relationships be used to predict the other, however the predictive between the species richness values for the five power is low with only 15.6% of the variance in one taxa examined. Therefore, none of these groups of taxa contributed by the variance of the other. In terrestrial invertebrates would be a reliable addition, this coefficient decreases slightly when “indicator” or “umbrella” group for any other, and controlling for the effect of other taxa (r = 0.352, P therefore are inappropriate for predicting overall = 0.018, n = 43). None of the other combinations of biodiversity at this scale. taxa have statistically significant correlations (see Given the effects of sampling “scale” on Table 3.24). species richness identified in Section 3.3.3.6, and To further examine the nature of the the influence of α- and β-diversity patterns relationship between richness values for ants and (Section 3.3.4.6), the richness data were re- spiders, the data were examined graphically analysed using the plot-based values (sum of 4 (Figure 3.31). It is obvious that the apparent sub-plots) for each treatment. Table 3.26 gives the positive correlation between these two taxa stems correlation coefficients (and probability values) for primarily from the disparity between richness the relationship between richness values for pairs values on burnt and unburnt sub-plots. Both taxa of taxa at the plot scale (n=6). The values in the have significantly higher values on burnt sub-plots bottom-left of the table represent correlation producing a relationship which is in fact an artefact coefficients for burnt sites, while those in the top- of this difference in richness. Similar patterns were right are coefficients for unburnt sites. There are exhibited for the relationship between flies and two statistically significant coefficients, ants &

241 Australia’s Biodiveristy - Responses to Fire

Table 3.25 Correlations between species richness values for five taxa (controlling for treatment). Values on the bottom-left represent standard Pearson’s Product-Moment Correlation Coefficients (n=24) for burnt sub-plots, those on the top-right are for unburnt sub-plots. Pairs of data represent correlation coefficient (top) and probability (bottom) values. UNBURNT ANTS BEETLES BUGS SPIDERS FLIES ANTS 0.042 0.086 0.064 -0.077 0.847 0.689 0.765 0.721 BEETLES 0.267 0.123 -0.213 0.386 0.208 0.568 0.318 0.063 BUGS 0.105 0.098 0.275 -0.035 0.624 0.648 0.193 0.871 SPIDERS 0.260 0.195 -0.181 0.175 0.220 0.361 0.397 0.414 FLIES 0.076 -0.104 0.024 0.101 0.724 0.628 0.912 0.640

Table 3.26 Correlations between species richness values for five taxa (controlling for treatment). Values on the bottom-left represent standard Pearson’s Product-Moment Correlation Coefficients (n=6) for burnt sub- plots, those on the top-right are for unburnt sub-plots. Pairs of data represent correlation coefficient (top) and probability (bottom) values. UNBURNT ANTS BEETLES BUGS SPIDERS FLIES ANTS -0.719 -0.968 0.603 0.353 0.107 0.001 0.205 0.492 BEETLES 0.751 0.664 -0.818 0.161 0.085 0.150 0.0460.760 BUGS 0.046 -0.185 -0.591 -0.388 0.930 0.7260.2160.447 SPIDERS 0.155 -0.189 -0.644 0.273 0.770 0.719 0.167 0.600 FLIES -0.120 -0.112 -0.021 0.186 0.821 0.832 0.968 0.724

bugs and beetles & spiders, but only on unburnt used to predict the other, and the predictive plots. None of the other combinations of taxa power is high with 66.9% of the variance in one have statistically significant correlations. taxa contributed by the variance of the other. On unburnt plots only, species richness These results suggest that, at least for these 2 values for ants and bugs are negatively correlated pairs of taxa, and at the 1ha sampling scale, it may (r= -0.968, P = 0.001, n = 6). This implies that as be possible and reliable to use one group as an the richness of ants increases, the richness of bugs indicator of the biodiversity of the other. decreases (and vice versa). One could be used to predict the other, and the predictive power is high with 93.7% of the variance in one taxa contributed by the variance of the other. Similarly, on unburnt plots only the species richness values for beetles and spiders are negatively correlated (r= -0.818, P = 0.046, n = 6). This implies that as the richness of beetles increases, the richness of spiders decreases (and vice versa). One could be

242 4. DISCUSSION

There is good theoretical and growing empirical 4.1 HABITAT STRUCTURE evidence to support the role of biodiversity in the maintenance of ecological processes within Low intensity fires used for fuel control generally forests. The multitude of organisms that result in incomplete combustion of surface litter constitute biodiversity play an essential role in and understorey vegetation (Tolhurst et al. 1992; primary production, nutrient cycling and uptake, Williams and Gill 1995). A mosaic of habitat population and community level interactions and patches results at a small scale, with these patches energy storage and transfer (see Majer 1992b; influencing the spatial distribution of surviving Woodward 1993; Beattie 1995). Through their terrestrial invertebrates and their ability to contribution to ecosystem function, these recolonise burnt areas. If repeated low-intensity organisms also enable forest ecosystems to fires reduce this spatial heterogeneity (see Fox and provide benefits to humanity. These include Fox 1986; Nieuwenhuis 1987) then this practice amenity values in the form of aesthetics, recreation may have long-term consequences for the survival and education; heritage values as forests contribute of invertebrate populations. In this study, 31 to long-term security for catchment protection, environmental variables were measured in order air & water quality and nature conservation; and to firstly, assess the long-term effect of frequent economic values including timber production, burning on the habitat, and secondly, to grazing and ecotourism (see Hobbs 1992; York investigate the nature of relationships between 1993; New 1995). habitat components and invertebrate biodiversity. The maintenance of biodiversity is a Four general trends with regard to fundamental principle underlying ecologically environmental variables were detected: large-scale sustainable management (NSESD 1992). State spatial patterns, site-dependent patterns, Forests of New South Wales, through its Corporate treatment-dependent patterns, and general Plan (1992), has stated that it will manage its patterns independent of site and treatment. At the forests on an ecologically sustainable basis using largest scale (the study area), there was a gradual best forest practices. Biodiversity conservation, N-S decrease in the mean biomass of leaves as a and hence ecological sustainability, cannot be component of the leaf litter. This pattern was only achieved without consideration of the important evident however on unburnt plots and role that invertebrates play, both through their disappeared with frequent burning, resulting in a involvement in ecological processes, and by their simplification of the large-scale spatial patterning substantial contribution to the overall richness of of this habitat component. Top-soil moisture biological communities. Invertebrates are the content tended to rise to a peak towards to centre most diverse and abundant animals in most of the N-S road transect and then decline towards natural ecosystems, but their importance in the southern end, with frequent burning having sustaining those systems is commonly not no effect on this spatial trend. appreciated (New 1995). Periodic low-intensity Frequent low-intensity fire had resulted in a fire (hazard-reduction burning) is a conspicuous number of treatment-related changes in measured management strategy in virtually all of Australia’s habitat parameters. Within the “fine-fuel” dry forest communities. While it is primarily used component, there was (on average) a 44% to reduce fuel levels, little is known about the reduction in leaf biomass, and a reduction in effects of its repeated use on natural ecosystems large-scale spatial trends (see above). There was over long time-scales. The primary objective of (on average) a 63% reduction in the very fine this study was therefore to assess whether the litter component, with the appearance of spatial frequent use of this forest management practice patterning not evident on unburnt plots. With the was compatible with the conservation of a major twig component, there had been (on average) a component of biodiversity: the terrestrial 50% reduction in the biomass of twigs 0-6mm, invertebrates. and a 44% reduction in the biomass of twigs 6- 25mm. Bark biomass had decreased (on average)

243 Australia’s Biodiveristy - Responses to Fire

by 36%, and increased in spatial heterogeneity. sized shrubs, the spatial homogeneity of certain These reductions in litter biomass largely reflect a litter components (biomass of leaves and very fine response to the most recent fuel-reduction burn material) and top-soil hardness (but see site- (2 years previously) and would be expected to dependent effects mentioned above). There was change with time as fuel continues to accumulate also no obvious change with regard to the (see Birk and Bridges 1989). Over the 20 year distribution of sticks & logs (>2.5cm) within period of frequent hazard-reduction burning, nominated size categories, however some changes litter biomass (fine fuel) had fluctuated between to the external nature of logs (charring) was 15 and 23 tonnes ha-1 on unburnt plots, and apparent. between 4 and 20 tonnes ha-1 on burnt plots There were a number of “general patterns” (York 1996). Burning removed between 46 and with regard to the response of environmental 73% of litter (by weight), but by 3 years post-fire variables to frequent fire. As the amount of litter levels had usually reached and often vegetation in the first metre above the ground exceeded that achieved before the previous fire increased, it became more spatially homogeneous (range = 72–136%). At the time of this study mean (less “patchy”). The amount of leaf litter (twigs, litter biomass was 9.3 tonnes ha-1 on burnt plots, bark, leaves and very fine material) at study sites representing about 50% of levels on unburnt was independent of the amount of understorey plots. vegetation, but was correlated (positively) with Other treatment-related effects involved the top-soil moisture levels and (negatively) with the amount (cover) of vegetation in the understorey amount and spatial variability of light reaching the and its spatial distribution. While the quantity of ground. This suggests that frequent fire has an vegetation in the layers closest to the ground independent influence on vegetation understorey (ground herbs and small shrubs) was not affected and leaf-litter habitats, and that (in this forest) the by frequent burning (see below), there was an leaf-litter environment exerts the primary control decrease in the spatial heterogeneity on top-soil moisture and surface insolation levels. (“patchiness”) of these layers. Conversely, there Christensen (1985) found that the removal of the was a substantial reduction in the cover of tall and litter layer by fire caused increased surface very tall shrubs (on average, 65% and 93% heating, leading to greater evaporation and lower respectively). Both these layers showed an increase moisture in the upper soil. This situation would in spatial heterogeneity with frequent burning. be expected to change with the time-related Top-soil moisture levels were, on average, 18% accumulation of litter after fire. lower following 20 years of frequent burning, While the changes to the amount of leaf whereas the amount of light reaching ground level litter, understorey vegetation and top-soil had increased (on average) by 125%, and moisture may only reflect a time-since-last-fire increased in its spatial heterogeneity. phenomenon, changes to the spatial variability Spatial patterning of a number of following frequent burning may reflect a decrease environmental variables was largely site- in habitat heterogeneity which, in turn, could dependent, with no overall large-scale or impact upon terrestrial invertebrate communities. treatment-related features. Top-soil hardness was There were measured reductions in the large- site-specific and independent of other variables. scale spatial patterning of the leaf litter, and There was a weak tendency for there to be greater changes to its physical structure with an increase amounts of vegetation in the first metre above in the patchiness of bark and very fine litter ground on sub-plots with more exposed (north- components. The increased patchiness of ground westerly) aspects, and there was a slight tendency insolation levels would appear primarily to be a for there to be a greater number of large logs on response to these changes. There was a decrease sub-plots with lower slope angles. in the spatial heterogeneity of ground herbs & A number of habitat components showed no small shrubs but an increase for tall & very tall significant response to frequent burning and did shrubs. This potentially reflects a change in the not appear to exhibit patterns that could be composition of the understorey vegetation attributed to large-scale spatial influences. These following frequent burning (see Fox and Fox were: the amount of vegetation in the first metre 1986; Nieuwenhuis 1987) and a removal of taller above the ground (ground herbs, small and mid- (older) shrubs due to the short interval between sized shrubs), the spatial homogeneity of mid- fires. The relationship of these environmental

244 Bushfire and forest invertebrates

patterns to the abundance and distribution of open habitats (CSIRO 1991) such as those typical terrestrial invertebrates is explored and discussed of frequently burnt areas, although Dolva (1993) in the following sections. found that wood crickets (Gryllidae) were more abundant in unburnt areas, probably in response 4.2 TERRESTRIAL INVERTEBRATE to the structure of the litter layer. Thrips are COMMUNITIES generally associated with vegetation and may be responding to the marked changes in the structure This study revealed a rich terrestrial invertebrate and cover of the understorey. fauna with representatives from the Chelicerata For the ten broad taxonomic groups where (spiders, ticks & mites, pseudoscorpions, there were sufficient data to permit statistical harvestmen), Crustacea (landhoppers, slaters), testing, the results indicated a variety of responses Chilopoda (centipedes), Diplopoda (millipedes), to frequent burning. Seven groups (isopods, and a large number of Insect Orders & Families. springtails, ticks & mites, bees & wasps, insect 4.2.1 Invertebrate Abundance larvae, flies and beetles) showed substantial decreases in abundance following frequent Numerically, the most abundant groups overall burning. These decreases ranged from 15 to 58% were the springtails (33.1%), ticks & mites (see Section 3.3.2), but were only statistically (23.9%), ants (23.1%), bugs (4.2%), beetles significant for ticks & mites (31%), insect larvae (4.0%), bees & wasps (2.8%), insect larvae (2.7%), (35%), flies (58%) and beetles (31%). Many of flies (2.6%) and spiders (2.2%), with these nine these groups are associated with leaf litter and it is groups making up 98.6% of the total number of likely that their numbers have been influenced by organisms caught. The first three groups the episodic removal of this resource, and the fact (springtails, ticks & mites, and ants) represented that litter levels on frequently burnt plots were, on 80% of individuals caught. average, 50% of that on unburnt plots. Newman Due to their low numbers, it was not and Tolhurst (1991) considered that reductions in possible to comment on the effects of frequent abundance of Collembola and Diptera following a burning for: pseudoscorpions, harvestmen, single fire event were in response to reduced litter centipedes, millipedes, diplurans, termites, (fuel) levels. Terrestrial mites are exceedingly embiids, booklice, lacewings, caddisflies, moths common in soil and leaf litter and are generally and butterflies. For these groups the trapping predatory, feeding on small invertebrates (Harvey method used may not have been the most and Yen 1989). Collembola communities have appropriate and has potentially contributed to the been shown to be initially particularly sensitive to low capture rate. While the low numbers fire (Campbell 1973), and an increase in fire collected for several other taxa precluded frequency may reduce population sizes and alter statistical analysis, frequent burning appears to community structure (Metz and Dindal 1975; have led to a reduction in the numbers of Dindal and Metz 1977). It was interesting to note amphipods, cockroaches and earwigs, and an that the large-scale patterns in abundance increase in the numbers of grasshoppers & exhibited by ticks & mites and the beetles crickets, and thrips. Terrestrial amphipods live in paralleled patterns in top-soil moisture content and feed on decaying litter of the forest floor, (see Figures 3.4, 3.8 and 3.14). The observed requiring a relatively moist environment because significant reduction in abundance of these groups they are susceptible to desiccation (Friend and (both down by 31%) may be in response to the Richardson 1986). Densities of some species have 18% decrease (on average) in top-soil moisture been correlated with litter thickness (eg., Duncan levels apparent on sites experiencing frequent fire. 1969) suggesting that the reduction in leaf litter High spatial variability in abundance for isopods, associated with hazard-reduction burning is likely springtails, and bees & wasps possibly contributed to result in lower population sizes for this group. to the lack of statistical significance in this study. Similarly, cockroaches feed mainly on the detritus Three groups showed substantial increases associated with leaf litter while the majority of in abundance following frequent burning. These earwigs feed on live or decaying plant matter were statistically significant for bugs (77%) and (Zborowski and Storey 1995). Both groups would ants (250%), but not for spiders (33%). This may be impacted upon by a reduction in the amount of be due to increased ease of movement (increased this resource. Surface active groups such as “trapability” - see Majer 1980, Andersen 1988) for grasshoppers and crickets tend to prefer more

245 Australia’s Biodiveristy - Responses to Fire

surface active groups such as spiders and ants, as some lacewings feeding on honeydew or pollen, well as changes in habitat suitability. The relative while others prey on aphids and scale-insects abundance of many groups was influenced not (Zborowski and Storey 1995). This group may be only by fire history, but also by site-specific more abundant in the more structurally complex habitat conditions (see below), with both spiders vegetation of unburnt sites, however a sampling and ants showing considerable spatial variability in regime concentrating on understorey vegetation their numbers. This suggested that large sample would be required to more accurately assess this. sizes would be required to detect management In order to test the utility of RBA effects when using coarse-scale taxonomic methodology, and further investigate the impact classification (eg. Family or Order). While the of repeated burning on species richness and the uncertain taxonomy of many groups precludes related aspects of community composition and finer scale resolution, recent developments in structure, five taxa were investigated in detail by Rapid Biodiversity Assessment (RBA) may permit analysis to morphospecies level. These groups subsets of the fauna to be investigated more fully utilise a diversity of micro-habitats and niches and (Oliver and Beattie 1993; Beattie and Oliver are representative of the range of terrestrial 1994). This would allow the use of smaller sample invertebrates found in these forest environments. sizes and produce a more cost-effective outcome These were: firstly, the Hemiptera (bugs), a (see below). mostly terrestrial and phytophagous (plant- feeding) group which have a close association with 4.2.2 Invertebrate Species Richness plant communities. Secondly, the Diptera (flies), Diversity (richness) at the Ordinal level varied which although highly mobile as adults, have from 11–17 broad taxa on individual study plots, particular requirements with regard to larval food with frequent burning significantly reducing sources; usually moist, decaying plant and animal diversity at this scale. While on average this material. Many species are parasitic on the larvae decrease was slight (≈ 1 Order per sub-plot), 4 of other insect orders with specialist habitat taxa were missing overall from frequently burnt requirements for oviposition. Thirdly, the plots. These were the Opilionida (harvestmen), Araneae (spiders), a major group of predators in Embioptera (embiids), Psocoptera (booklice) and forest ecosystems exploiting a variety of habitats. Neuroptera (lacewings). The Opilionida They live in burrows or crevices in the ground, (harvestmen) are small to medium Arachnids amongst leaf litter or in vegetation, and are a (<10mm body length) which are usually found in group with many habitat specialists. Fourthly, the moist leaf litter, or under rocks, logs and bark. Coleoptera (beetles), which utilise a diverse range Most feed on smaller invertebrates but some also of habitats & micro-habitats, with a variety of consume plant material (Harvey and Yen 1989). feeding strategies (adults include herbivores, Given their habitat requirements, it is expected predators & scavengers, while larval forms feed that they would be disadvantaged by the drier either internally or externally on plants and fungal conditions found on frequently burnt areas. products). Beetles are a rich and diverse group Embiids (web-spinners) are small to medium which are active in the litter layer. Lastly, the (4–15mm body length) insects usually living under Formicidae (ants), which are one of the most rocks, bark or leaf litter. They feed on leaves, numerous and widespread groups in Australian bark, mosses and lichens (Zborowski and Storey ecosystems. They have a diverse diet, and utilise a 1995). The Psocoptera (booklice, psocids) are variety of feeding strategies from predators and minute to small (<1–10mm body length) insects scavengers, to plant eaters and fungus feeders, which live on vegetation, or under bark or stones. with frequent and varied interactions with other They feed on minute organic items such as plant invertebrate groups. Ants nest in the soil and litter spores, algae, lichen and fungi. For both groups, and therefore are response to disturbance of these these food resources are likely to be more habitats, and they are functionally important abundant in the moister litter and soil conditions within the forest ecosystem. prevalent in infrequently burnt forest. The Overall, 411 morphospecies were identified Neuroptera (lacewings) are small to large from the five groups studied in detail. The beetles (wingspan 5–150mm) insects with generally were the most species rich (139 morphospecies), active, long-legged predacious larvae (antlions). followed by the ants (88), flies (77), spiders (63), Adults may be predacious or omnivorous, with and bugs (44). The results of analyses (ANOVA)

246 Bushfire and forest invertebrates

investigating the effects of frequent burning and observation is the lack of concordance in spatial patterns due to large-scale spatial effects indicated patterns of richness between taxonomic groups. a variety of responses to frequent burning. Two This suggests that groups are responding groups, flies and beetles, experienced a significant differently to environmental factors and to the reduction in species richness on sub-plots agent of disturbance (frequent fire). This has following frequent burning (44% and 27% implications not only for invertebrate sampling reduction respectively). A further two groups, the strategies, but also for the use of a single, or bugs and the spiders, showed an increase in limited group of taxa as a surrogate in biodiversity species richness on sub-plots (16% and 27% assessment. This supports the findings of Oliver respectively), although these results were not (1995) who found that different taxa responded to statistically significant. The ants experienced a the disturbance of forest logging in distinct ways, significant increase in sub-plot richness (26%) and that it was not appropriate to use any one following repeated burning. Few studies have taxon as a surrogate for the richness of any others dealt with the impact of fire on invertebrate in conservation evaluation, environmental species richness. Leonard (1972) found that the monitoring or impact assessment. species richness of leaf litter fauna may drop by It was apparent that estimates of species 50% immediately after fire. Recovery after fire richness were also influenced by the spatial scale may be rapid (Leonard 1972), or take several years of measurement, with associated implications for (Moulton 1982), depending upon the season of the interpretation of observed treatment effects burn and the meteorological conditions following for the different taxa. For bugs and flies results the fire and the influence these factors have on were consistent across a range of scales of litter accumulation, and the mobility and measurement, with the magnitude and direction recolonising ability of particular species (Morris of differences between unburnt and burnt results 1975). Long-term studies of spiders (Huhta 1971; for sub-plot, plot and treatment similar. For Merrett 1976) and ants (Brian et al. 1976; York spiders, while the magnitude and direction of 1994) have shown a replacement of species in the species richness at the scale of sub-plot and plot years after fire which is related to their particular were similar, considerably more species were habitat requirements being met as the vegetation found overall on burnt compared to unburnt structure changes over time. Species richness may plots. This suggested that species assemblages on stay largely unchanged (Merrett 1976) or decline burnt plots were more diverse than those on (York 1994). unburnt plots, resulting in higher β- (between- At the scale at which richness (α-diversity) habitat) diversity. Diversity on sub-plots within was estimated in this study there were large-scale both unburnt and burnt plots would appear to be spatial patterns exhibited by some groups. For similar, suggesting the differences lie at, or above, bugs, spiders, beetles and ants these were not the scale of plot ( 1 hectare). For beetles, the statistically significant overall, but nevertheless magnitude of the difference detected between have implications for future projects attempting to unburnt and burnt at the scale of sub-plot and measure disturbance impacts. For ants, observed plot were similar, however the direction was spatial variation in richness appeared to reflect reversed at the scale of treatment. This would random variation expected within a sampling suggest a similar situation as to that with the program such as this. For other groups observed spiders, where the species assemblages on burnt variation was distinctly non-random and appeared plots are more diverse than those on unburnt to reflect underlying environmental patterns. For plots. Diversity on sub-plots within both unburnt bugs similar trends in richness were apparent at and burnt plots would appear to be similar, paired sites along the road transect suggesting that suggesting the differences lie at, or above, the a smaller number of replicates would have been scale of plot ( 1 hectare). For ants, the magnitude sufficient to detect impact (or lack of impact). For of the difference in species richness detected flies, spiders and beetles this was not the case. between unburnt and burnt areas at the scale of Exhibited spatial patterns were often quite sub-plot and plot were similar, although the different between the two treatments for a single magnitude of the difference was reduced at the taxa (eg. Flies - Section 3.3.3.2), reinforcing the scale of plot (compared with other taxa). The need for sufficient experimental replication in direction of the difference was however order to detect real differences. A most important substantially reversed at the scale of treatment,

247 Australia’s Biodiveristy - Responses to Fire

suggesting a different situation to that with the proportions of morphospecies in each category spiders and beetles, with the species assemblages however varied substantially between taxonomic on burnt plots less diverse. Diversity on sub-plots groups. For Hemiptera (bugs) the proportions within both unburnt and burnt plots would appear were 16, 41 and 43% for both, unburnt and burnt to be less similar than with other taxa, suggesting respectively; for Diptera (flies) 45, 40 and 15%; the differences lie at less than the scale of plot ( 1 for spiders 25, 24 and 51%; for beetles 28, 34 and hectare). 38%; and for ants 57, 23 and 20%. The overall These spatial patterns in estimates of species biodiversity of frequently burnt areas was richness are a consequence of the composition of maintained by the addition of species not invertebrate assemblages (communities), and their recorded on unburnt plots. The changed response to habitat conditions, at the varying environment was supporting an additional 133 scales of investigation. The nature of these morphospecies (19 bug, 11 fly, 32 spider, 53 beetle patterns, and their interaction with environmental and 18 ant species). It is notable that a large variables, will be further explored in the following proportion of species (16-47%) are apparently section. indifferent to disturbance history and habitat structure within the limits sampled in this survey, 4.2.3 Community Composition although changes to relative abundance need to be The five groups studies in detail proved to be taken into consideration. extremely diverse. Beetles had the richness fauna Although an examination of relative overall with 139 beetle morphospecies abundance patterns enables broad “assemblages” representative of nine super-families and 25 of species with similar responses to disturbance to families. The ants were the second richness group be identified, these different patterns were more with 88 morphospecies representative of 5 sub- clearly apparent from a comparison of bi-plots families and 34 genera. They were followed by the derived from the CCA ordination procedure. For flies with 77 morphospecies representative of 2 bugs, spiders, beetles and ants there was little or sub-orders and 20 families, the spiders with 63 no overlap of unburnt and burnt sub-plots in morphospecies representative of 21 families, and ordination space, indicating low similarity of the the bugs with 44 morphospecies representative of species assemblages of the two treatments. For 16 family (or similar) groups. flies however there was a substantial overlap, Overall, the same number of morphospecies reflecting the relatively large number of (279) were collected from unburnt and burnt morphospecies shared by the two treatments. plots, with average (mean) richness values similar Morphospecies found in both burnt and unburnt on both treatments (48.2 and 46.5 morphospecies areas can be regarded as habitat “generalists”, respectively). This initially suggests that frequent largely resilient to frequent disturbance and burning had not reduced biodiversity in this forest therefore of lesser concern with regard to environment. An analysis of the richness (α- and biodiversity conservation. Of greater importance β-diversity) of individual taxonomic groups has are those species absent from frequently burnt shown this not to be the case, with groups sites (potential habitat “specialists”). The results responding differently to frequent burning. The from this study suggest that frequent burning had nature and, potentially, the mechanisms behind led to the loss of up to 131 species (18 bugs, 31 this difference can only be elucidated by an flies, 15 spiders, 47 beetles and 20 ants), which examination of the species composition of faunal represents 47% of the morphospecies known from assemblages (communities) and would not be the unburnt areas. Many of the morphospecies apparent from an examination of data at a higher apparently lost from frequently burnt sites were taxonomic level (for example: Neumann and however only detected on a single sub-plot or Tolhurst 1991; Neumann 1992; Coy 1996). represented by a single individual on unburnt An inspection of the distribution of plots. These could be genuinely rare or morphospecies across sub-plots for each treatment uncommon species which were missed purely by detected a consistent pattern, irrespective of chance when sampling burnt plots. For this reason faunal group. Morphospecies fell into one of three it is difficult to identify clear patterns (at the groups: found on both treatments (Group A), species level) from the relative abundance data found only on unburnt plots (Group B), or found alone. Some general trends were apparent only on burnt plots (Group C). The relative however when morphospecies data were arranged

248 Bushfire and forest invertebrates

by genus (ants) or family (bugs, flies, spiders and following frequent burning, and lower top-soil beetles) utilising general information available on moisture levels, would appear to offer an their biology and ecology at these taxonomic explanation to the reduction in the number of levels (“guild” / “functional group” approach). species from the families Sciaridae, Phoridae and For bugs, unburnt plots had more species Empididae. Results of the CCA ordination from the infra-order Dipsocoromorpha (7 vs 3), support these conclusions with unburnt sub-plots while burnt plots have greater numbers of species characterised by high levels of litter, high cover of from the family Reduviidae (5 vs 0). The tall and very tall shrubs, high top-soil moisture Dipsocoromorpha include species (described & levels and low and spatially variable amounts of undescribed) primarily known from leaf litter and insolation at ground level. Burnt sub-plots were other moist environments (CSIRO 1991). The characterised by high levels of insolation at observed substantially lower (50%) amounts of ground level, greater exposure (more north- leaf litter and changes to its spatial distribution westerly aspects), and to a lesser extent, steeper following frequent burning would appear to offer slopes. These drier, more exposed conditions an explanation to the reduction in would be less favourable for species from these fly Dipsocoromorpha species. Results of the CCA families. This supports the findings of Delettre ordination support these conclusions with (1994) who, in a study of a heathland chironomid unburnt sub-plots characterised by high levels of (midge) community, found that species litter, high cover of tall and very tall shrubs, high composition was best explained by fire-related top-soil moisture levels and low and spatially changes to the vegetation structure and soil variable amounts of insolation at ground level. moisture levels. Burnt sub-plots were characterised by high levels For spiders, unburnt plots had more species of insolation at ground level, greater top-soil from the family Malkaridae (2 vs 0), while burnt hardness and greater cover of the herb & shrub plots had greater numbers of species from the component of the understorey vegetation. The families Zodariidae (9 vs 2), Gnaphosidae (6 vs 3), Reduviidae (Assassin Bugs) are predacious on Corinnidae (7 vs 4), Linyphiidae (4 vs 2) and other invertebrates and found generally on Lycosidae (3 vs 0). The Malkaridae are moist vegetation and on the ground. There is no clear habitat specialists, commonly dwelling in the leaf explanation as to their absence from unburnt litter (M.Gray pers. com.). The Lycosidae (wolf forest and they may be useful disturbance spiders) and Zorariidae are small to large, ground “indicators” (see 4.2.5). living, hunting spiders. Species from the For flies, unburnt plots had more species Zorariidae are vagrant hunters, frequenting areas from the families Sciaridae (8 vs 5), Phoridae (16 with an open vegetation structure and low litter vs 13) and Empididae (6 vs 3). The Sciaridae levels. The Gnaphosidae and Corinnidae are both (black fungus gnats) are often associated with generalised hunters, tolerant of drier conditions. decaying material, with their larvae are often The Linyphiidae (tent spiders) are good found in rotting vegetable matter or highly colonisers of disturbed habitats and can “balloon- organic soils (CSIRO 1991). The Phoridae in” from shrubs some 50m distant (M.Gray pers. (humpbacked flies) are active scavengers on com.). The observed substantially lower amounts foliage and litter, with the larvae generally of leaf litter and changes to its spatial distribution scavengers in carrion and other decomposing following frequent burning, and lower top-soil matter. Adults generally oviposit in carrion and moisture levels, would appear to offer an organic material on the ground (D.Bickel pers. explanation to the reduction in the number of com.) All the morphospecies of Empididae found species from the family Malkaridae, and the in this study were from the sub-family increase in species from the families Zodariidae, Tachydromiinae, which are mainly terrestrial and Gnaphosidae, Corinnidae, Linyphiidae and rarely fly. As adults they are generally predacious Lycosidae. Results of the CCA ordination support on smaller arthropods and frequent moist places, these conclusions with unburnt sub-plots commonly amongst vegetation. The larvae are characterised by high levels of litter, high cover of probably predacious, living in the soil and within tall and very tall shrubs, high top-soil moisture decaying vegetation and the leaf litter (CSIRO levels and low and spatially variable amounts of 1991). The observed substantially lower amounts insolation at ground level. Similarly, burnt sub- of leaf litter and changes to its spatial distribution plots were characterised by high levels of

249 Australia’s Biodiveristy - Responses to Fire

insolation at ground level and greater cover of the following frequent burning, and lower top-soil herb & shrub component of the understorey moisture levels, would appear to offer an vegetation. As a number of Corinnidae species are explanation to the reduction in the number of ant specialists, the substantial increase in ant species from the genera Cerapachys and abundance on frequently burnt sites may have also Hypoponera. Similarly, the structurally simplified contributed to the increased richness of this environment found on frequently burnt areas family through increased prey availability. would provide suitable habitats for large, solitary For beetles, unburnt plots had more species foragers such as Colobostruma spp. Results of the from the family Carabidae (11 vs 8), while burnt CCA ordination support these conclusions with plots had greater numbers of species from the unburnt sub-plots characterised by high levels of families Curculionidae (18 vs 9) and litter, high cover of tall and very tall shrubs, high Chrysomelidae (8 vs 4). The Carabidae (Ground top-soil moisture levels and low and spatially Beetles) are mainly predatory, both as adults and variable amounts of insolation at ground level. larvae, on plant-inhabiting insects. The Similarly, burnt sub-plots were characterised by Curculionidae (Weevils), as adults, feed on the high levels of insolation at ground level and stems, roots, seeds and fruits of plants, with larvae greater cover of the herb & shrub component of usually feeding on wood and other plant parts. the understorey vegetation. The Chrysomelidae (Leaf Beetles) feed on leaves Across the five groups studied in detail; bugs, and other (living) vegetative parts of plants, both flies, spiders beetles and ants, there was a as larvae and adults. The richness of the beetle consistent pattern with regard to the changes that fauna here would appear to reduce the ability to occur to community composition following generalise at the level of family. Results of the frequent burning. Although species richness (α- CCA ordination suggest a reasonable separation diversity) decreased by 44% and 27% for flies and of assemblages on burnt and unburnt plots, with beetles, and increased by 16%, 27% and 26% for unburnt sub-plots characterised by high levels of bugs, spiders and ants (respectively), all groups litter, high cover of tall and very tall shrubs, high experienced a loss of species with frequent top-soil moisture levels and low and spatially burning. The results from this study suggest that variable amounts of insolation at ground level. frequent burning had led to the loss of up to 131 Similarly, burnt sub-plots were characterised by species (18 bugs, 31 flies, 15 spiders, 47 beetles high levels of insolation at ground level. It is and 20 ants), which represents 47% of the reasonable to suggest that ground-dwelling morphospecies known from the unburnt areas. species from the Carabidae are influenced by The losses were disproportionate across the substantially lower amounts of leaf litter and groups with percentage reductions ranging from changes to its spatial distribution following 41% (bugs), to 40% (flies), 34% (beetles), 24% frequent burning. Changes to the structure of the (spiders) and 23% (ants). The species lost would vegetation community with frequent burning appear to be from those groups dependent upon a appear to have provided additional habitats for substantial litter layer and stable moist conditions. plant-dependent species from the families The overall biodiversity of frequently burnt areas Curculionidae and Chrysomelidae. was maintained by the addition of species not For ants, unburnt plots had more species recorded on unburnt plots. The changed from the genera Cerapachys (7 vs 1) and Hypoponera environment was supporting an additional 133 (4 vs 1), while burnt plots had greater numbers of morphospecies (19 bugs, 11 flies, 32 spiders, 53 species from the genus Colobostruma (2 vs 0). beetles and 18 ants). These species would appear Species of Cerapachys are considered by Andersen to have broad tolerances, or adaptations, to drier (1990) to be “climate specialists” and are known to and more open environments. Overall, the be specialist predators, often on other ants composition of terrestrial invertebrate (Holldobler and Wilson 1990). Species from the communities was therefore influenced by a genus Hypoponera are cryptic, nesting and foraging combination of site-dependent (slope and aspect) within the soil and leaf litter. Members of the and treatment-dependent (litter, insolation, herb genus Colobostruma are large solitary & shrub cover, top-soil moisture & hardness) foragers/specialist predators (Andersen 1990). environmental variables. The observed substantially lower amounts of leaf This research has also shown that patterns of litter and changes to its spatial distribution relative abundance and species richness are not

250 Bushfire and forest invertebrates

concordant between broad taxa. Analysis of increased for spiders, but appeared unchanged for community composition, as illustrated by the flies. For bugs, beetles and ants it would appear CCA ordination, also illustrates that patterns of that the degree of heterogeneity of the vegetation invertebrate community organisation show varied near the ground plays a role in maintaining responses to environmental disturbance. In the bi- biodiversity, probably by providing additional plots derived from the CCA ordination, the habitats and an associated increase in food and degree of clustering of the sub-plots from each other resources. Structural heterogeneity may also treatment indicates the relative similarity of imply a greater diversity of plant species, or of species assemblages on sub-plots and plots within growth stages of existing species. Increased levels each treatment. The tighter clustering of burnt of shading and changes in surface and nest sub-plots for bugs, beetles and ants indicated a temperature with increased vegetation cover has lower within-treatment diversity compared to been shown to reduce ant species richness unburnt sub-plots (ie. a lower β-diversity). The (Goldstein 1975, Greenslade and Mott 1979), and converse applied for spiders, with the tighter alter community composition as the clustering of unburnt sub-plots indicating a lower environmental conditions become sub-optimal for β-diversity compared to burnt sub-plots. The certain species (Welch 1978, Elmes and Wardlaw situation for flies indicated similar within- 1982, York 1994). Moisture, light and temperature treatment diversity for both treatments, with loose have been identified as important factors in the clustering of both unburnt and burnt sub-plots. determination of the composition of invertebrate This interaction between point richness (α- assemblages (Huhta et al. 1967, Punttila et al. diversity) and spatial “turnover” of species (β- 1991, McIver et al. 1992). Frequent burning led to diversity) has substantial implications for the greater spatial heterogeneity in the bark and twig interpretation of the apparent effect of repeated components of the litter layer, with a likely impact burning (see “scale effects” in Section 3.3.3.6). on its structural complexity. The composition of Similar patterns were apparent for bugs and ants spider communities has previously been shown to where richness values (α-diversity) on burnt sub- be influenced by the structure of the litter layer plots were on average higher than on unburnt following fire (Huhta 1971). The implication of sub-plots, however the high similarity of these changes for community organisation and assemblages on burnt sub-plots (low β-diversity) ecosystem function are considered in the meant that the overall richness of both treatments following section. were similar. Unburnt sub-plots had lower richness (α-diversity) but are less similar, resulting 4.2.4 Community Structure in higher “turnover” between sub-plots (higher β- The biological structure of a community involves diversity), increasing overall species richness for species composition and abundance, temporal that treatment. For flies and beetles burnt plots changes in communities, and the relationships had lower richness (α-diversity) but were less between species in communities. This in turn similar, resulting in higher β-diversity. Spiders exerts strong influences on the functioning of the exhibited a different pattern with both higher α- community, in other words, how the community and β-diversity for the burnt treatment, resulting works as a processor of energy and nutrients in a large number of species (32) unique to (Krebs 1985). The continued functioning of frequently burnt sites. communities and their ecological processes is a An explanation for the different levels of primary goal of ecologically sustainable species turnover (β-diversity) for the different management (ESDWG-Forest Use 1991). groups is likely to be found in patterns of While it is possible to describe and assess environmental heterogeneity and its effect on communities using indices such as species richness, micro-habitat diversity. Frequently burnt areas or to compare the relative abundance of species were shown to have more spatially homogeneous using similarity indices, multi-variate approaches ground herb, small shrub, tall and very tall shrub and/or through graphical representation, these layers, more spatially heterogeneous levels of provide little information concerning the processes insolation at ground level, and more spatially underlying these differences or any indication as to heterogeneous bark and twig components of the the relative sensitivity of species to landscape litter layer. The β-diversity of bug, beetle and ant change (Samways 1994). In order to simplify and communities was reduced with frequent burning, interpret the complexity of ecological systems, one

251 Australia’s Biodiveristy - Responses to Fire

approach has been to group species into “guilds” dung. Among the beetles, the Leiodidae are or “functional groups”. These groups recognise abundant in decaying organic matter and also the ecological rather than the taxonomic affinity of occur in carrion and fungal fruiting bodies. Many species. To test the applicability of this approach, are general scavengers, but certain groups are morphospecies (see Oliver and Beattie 1993) were associated with particular fungi (Lawrence and allocated to guilds based upon their feeding Britton 1994). The Ptiliidae are minute beetles strategies and habitat preferences, following which are relatively abundant in decaying organic reference to the relevant literature and discussions matter, including leaf litter, where their major with taxonomic experts. By classifying large food source appears to be fungal spores and numbers of species into smaller, more manageable hyphae. The Endomychidae feed on a variety of groups, it is possible to substantially reduce the fungi, with many occurring in leaf litter in moist apparent complexity of ecological systems and habitats. provide a basis for evaluating environmental For flies, the number of generalists and change (Andersen 1990). scavengers decreased slightly (11%), while for For all groups studied, frequent burning beetles the proportion of generalists increased, on resulted in a change in the structure of the average, by 79%, primarily through an increase in community. With regard to feeding strategy, there morphospecies from the family Scarabaeidae. was on average, a 15, 140 and 250% increase Scarab beetles always live in concealed habitats, (respectively) in the number of phytophagous feeding on roots, dung or decaying vegetable (plant feeding) species of bugs, flies and beetles. matter (McQuillan 1985). The reason for their With regard to the proportion of predator species, dramatic increase here is not apparent, although it remained unchanged for flies and bugs, however with their generally large size, their mobility may for bugs there was a total shift from the family be enhanced in the more open environment of Nabidae to the family Reduviidae. The Nabidae frequently burnt areas. With ants, the proportion are a family of predacious bugs whose eggs are of generalist morphospecies decreased, on oviposited into grass stems (CSIRO 1991). This average, by 11% following frequent burning. suggests that a change in this component of the Results were not consistent within the group with vegetation may be influencing the suitability of the genera Pheidole and Crematogaster decreasing the habitat for these species, which are then by 35 and 19% respectively, and the genus replaced by another group of predators from a Monomorium increasing by 388%. In these dry different family. With beetles there was, on forests, species from the genus Monomorium are average, an 18% decrease in the number of major seed predators (Andersen 1985, Andersen predator species, due primarily to a 54% and Ashton 1985), and may be responding to reduction in the number of morphospecies from changes in the vegetation composition that the family Staphylinidae. Most species from this accompany a frequent fire regime (Zedler et al. family are small, often cryptic and live as 1983, Nieuwenhuis 1987, Cary and Morrison predators hidden in soil and leaf litter. Others are 1995). For flies, the number of morphospecies associated with dung, carrion or fungi (Zborowski regarded as wide-ranging “tourists” was, on and Storey 1995). It is anticipated that the average, reduced by 43% on frequently burnt reduction in litter levels and decrease in soil plots. moisture associated with frequent burning would When groups were compared with regard to disadvantage this group. habitat preferences, it was apparent that there For groups feeding primarily on fungal were substantial changes in community structure products there was, on average, a 44% (flies) and for some taxa. For bugs, numbers of moist habitat 31% (beetles) decrease in the number of specialists from the sub-order Dipsocoromorpha morphospecies. For flies this was most marked in were, on average, reduced by 83%. Amongst the families Scaridae and Scatopsidae, and for spiders, moist habitat specialists were reduced by beetles, in the families Leiodidae, Ptiliidae & 88%, primarily due to a 90-95% decrease in Endomychidae. Species from the family Scaridae numbers of morphospecies from the families oviposit in fungi in the soil and also feed on fungal Theridiidae, Toxopidae and Oonopidae. The products, while the Scatopsidae are generally Theridiidae build their webs in leaf litter, the found in moist forest environments with their Toxopidae are a moist-adapted group, and the larvae occurring in rotting vegetable matter and Oonopidae mostly inhabit the litter in moist

252 Bushfire and forest invertebrates

environments. With the flies, the number of moist there was a substantial increase overall (30%) in habitat specialists remained similar, although the the average proportion of “opportunist” family Ceratopogonidae was more commonly morphospecies, due primarily to a single species: represented on burnt sub-plots and the family Rhytidoponera metallica, which was extremely Chironomidae on unburnt sub-plots. The numerous on burnt sub-plots. R. metallica is a well proportion of temperature-dependent (climate) known coloniser of disturbed habitats (Yeatman specialists among the ants remained largely stable and Greenslade 1980). (increased by 4%), although individual genera did It has been observed that the structure of ant vary in their response. For spiders, there was, on communities, in particular, may be influenced by average, and an 35% increase in the number of the relative abundance of particular “dominant” species with a known preference for dry habitats. and “sub-dominant’ groups (Fox & Fox 1982, For groups primarily inhabiting the litter Andersen 1990). Following frequent fire, the layer there was a variety of responses to frequent number of morphospecies within the “dominant” burning. For flies, the number of morphospecies functional group had, on average, decreased by specifically utilising the litter layer had, on 23%, primarily through a reduction in the average, decreased by 60%. This was primarily occurrence of Iridomyrmex species. The number of due to the absence of the family Tipulidae on “sub-dominants” had increased slightly (25%), frequently burnt sub-plots. Tipulidae (crane flies) largely because of an increase in the occurrence of use moist soil for breeding, with their larvae Camponotus species. The number of larger common in decaying vegetation, and therefore “solitary/specialist” morphospecies, which interact would be disadvantaged by frequent burning. only slightly with other groups, had increased With ants there was a 19% decrease in the slightly (13%), although the results were quite average number of “cryptic” morphospecies variable within the group. Three new genera were inhabiting the litter and soil, particularly with found on burnt plots (Colobostruma, Epopostruma & regard to the genera Hypoponera (90%) and Leptogenys) while the numbers of species of Tapinoma (86%). Species from the genus Myrmecia increased, on average, by 87%. Hypoponera are specialist predators, feeding largely Myrmecia have been shown to be more common on Collembola which were approximately 15% in recently burnt habitats in these forests (York less abundant on frequently burnt sites. Tapinoma 1994, 1996). The occurrence of a solitary forager, species are cryptic omnivores in the litter, with Bothroponera sp.A, had decreased substantially some arboreal nesters. Habitat availability for this (76%). Bothroponera are predacious, often on genus would be reduced by frequent burning. For termites (Holldobler and Wilson 1990), a group spiders however, the number of litter dwelling of organisms not found on frequently burnt sites species increased, on average, by 110% with an in this study. While a number of termite mounds equivalent increase from the families Hahniidae were observed on these areas, they did not appear and Textricellidae. Species from the family to be active. Hahniidae are small spiders that construct small sheet webs in litter and foliage. It would appear 4.2.5 Biodiversity Indicators that they are not disadvantaged by lower litter Programs in land appraisal and applied resource levels (M.Gray pers. com.). Spiders in the family management increasingly utilise "environmental Textricellidae are also very small and live deep in indicators" to facilitate and simplify assessment the litter layer near the litter/soil interface. As and decision-making procedures. Indicators may only 46-73% of litter is removed in each fire event take the form of an index which concisely in these forests (York 1996), this group may not be summarises some property of the system, such as disadvantaged by frequent fire. abundance or species richness (diversity); or Many groups of terrestrial invertebrates are describes the community via its species adapted to exploit disturbed habitats. For spider composition, the relative abundance of individuals morphospecies known to prefer open and within constituent species (evenness etc), or its disturbed habitats, the number of morphospecies organisation or “structure” (eg. number of guilds increased, on average, by over 600% following or functional groups). An alternative (or frequent burning, This was due primarily to the complimentary) approach may be to use occurrence of seven species from the family “indicator taxa”: an organism (or group of Zodariidae only on burnt sub-plots. For the ants organisms) that reveals important aspects of the

253 Australia’s Biodiveristy - Responses to Fire

structure and function for some part of the assessment of the impact of fire (see Campbell and ecosystem without exhaustive study of that part Tanton 1981; Majer 1984; Friend 1996). Temporal (Cornaby 1977). variability following single fires has been shown to A wide range of terrestrial invertebrates have be substantial, with taxa responding to seasonal and been used as “indicators” in Australia. Examples meteorological cues (see Neumann 1992; Coy include spiders (Mawson 1986), springtails 1996). Where sampling is required for comparative (Greenslade 1984, 1985; Greenslade and purposes only, for example burnt/unburnt, Greenslade 1987), termites (Nichols and Bunn logged/unlogged, rehabilitated/not rehabilitated 1980; Greenslade 1985), beetles (Greenslade contrasts, then the influence of temporal variability 1985; Yen 1987) and ants (Weir 1978; Majer can be reduced (controlled for) by simultaneous 1980b, 1984, 1985; Whelan et al. 1980; Yeatman sampling in the various treatment categories under and Greenslade 1980; Majer et al. 1982, 1984; examination (see Yeatman and Greenslade 1980; Andersen and McKaige 1987). Arthropods have a Majer et al. 1984; Burbidge et al. 1992). York significant role in the forest community, affecting (1994) showed that estimates of ant species richness both primary production by their grazing from a single summer sample of chronosequence activities, and the turnover of nutrients in their sites provided an accurate representation of long- role as decomposers (Lowman 1982). Forest floor term changes over time at a single site. Oliver and arthropods regulate microfloral decomposer Beattie (1996b) demonstrated that the richness of populations by their feeding on bacteria and ants recorded from a single summer pitfall sample fungal colonies, their transport of spores, and by was significantly correlated with richness values the contribution of their faeces and bodies for assessed by other sampling methods and seasons of decomposition (Van der Drift 1958, Engelmann sampling. This research therefore utilised a single 1961, MacFadyen 1962). Their requirements of a summer sample to assess the implications of spatial source of cover and food has led arthropods to variation in α- and β-diversity for investigations develop a sensitivity and responsiveness to system into frequent disturbance. structure which has made them useful indicators In this study, estimates of abundance were of system status and condition (Mattson 1977). shown to be influenced, not only by fire This research used two complimentary (treatment), but by large-scale spatial effects approaches to investigate the applicability of (position), with frequent interactions between invertebrate indicators for environmental impact these two factors. This meant that the particular assessment. This was undertaken firstly by looking fire effect (or lack thereof) was not always at spatial characteristics of commonly used consistent across the spatial range of sites biodiversity indices (abundance & richness) and (replicates). These interactions were significant (ie. by examining the concordance of these indices not as a consequence of random variation) for between taxa. It is often postulated that one group spiders, isopods (slaters), flies, beetles and ants. At of invertebrates may act as an indicator or times spatial patterns were evident, potentially in “umbrella” group for others, thereby allowing response to underlying environmental trends. This inferences to be made on the impact of was the case for ticks & mites and beetles. disturbance regimes. This assertion however relies Similarly, estimates of richness (α-diversity) were on as yet untested assumptions which flow from shown to be influenced by large-scale spatial particular sampling strategies, in particular, spatial effects (position) for flies and spiders, and patterns in species richness (α-diversity) and interactions between treatment and position effects species turnover (β-diversity). The second (spiders, beetles and ants). These patterns, approach involved a study of the composition and combined with the often considerable spatial structure of these communities in order to variability exhibited within a single treatment, identify species, groups of species, or community suggests that large sample sizes would be required “descriptors” which were useful as a means of to detect disturbance effects when using coarse- assessing potential effects on ecosystem function. scale taxonomic classification (eg Family or This was undertaken to determine the Order). Neumann (1992) recognised that while applicability of these methods for assessing such a broad taxonomic approach gives a ecological sustainability. cumulative estimate of the responses of the species High levels of spatial and temporal variability within each taxon, the behaviour of individual in invertebrate populations severely complicate an species remain unknown, thereby limiting the level

254 Bushfire and forest invertebrates

of interpretation that could be applied to the sustainable management. When differences in results. From an analysis of species-level data, this composition between burnt and unburnt areas research has shown those limitations to be were compared, it was apparent that frequent substantial when dealing with biodiversity issues. burning had led to a marked decline in species The concordance between the species with particular habitat preferences. The groups richness (α-diversity) of selected taxa was which were indicative of this decline were the investigated here using correlation analyses. infra-order Dipsocoromorpha and the families While a number of statistically significant Sciaridae, Phoridae and Empidae (flies), the correlations were detected, the predictive power families Malkaridae (spiders), Carabidae (beetles), of these relationships was weak. In addition it was and for ants, the genera Cerapachys and found that apparently significant correlations Hypoponera. Conversely, a number of groups were detected when using the whole dataset (spiders & “disturbance indicators” through marked increases ants - positive, flies & ants - negative, flies & in their richness on frequently burnt areas. These beetles - positive) did not hold when the two were the families Reduvidae (bugs), Zorariidae, treatments (burnt and unburnt) were analysed Gnaphosidae, Corinnidae, Linyphiidae and separately. It was evident that the apparent Lycosidae (spiders), Curculionidae and correlations between these taxa were an artefact of Chrysomelidae (beetles), and for ants, the genus the differences between average richness values on Colobostruma. The habitat parameters most burnt and unburnt sub-plots. Therefore, none of consistently associated with these changes were these groups of terrestrial invertebrates would be the amount and spatial distribution of leaf litter, a reliable “indicator” or “umbrella” group for any top-soil moisture levels, and the amount and other, and therefore are inappropriate for spatial heterogeneity of insolation levels at the predicting overall biodiversity at this scale. This ground surface. These environmental factors were supports the findings of Oliver and Beattie shown to vary primarily as a response to frequent (1996b) who demonstrated non-concordance burning, that is, under management control. between richness estimates for ants, beetles and Accompanying these habitat-related changes spiders. in community composition were a number of Given the identified influence of sampling shifts in ecological structure and function. For “scale” and spatial patterns of α- & β-diversity on some taxa the guild structure was generally estimates of species richness (see 4.2.2), maintained, but with changes to the suite of concordance between biodiversity estimates for species comprising each guild. This was most selected taxa were examined at a larger scale (1 marked with the ants, where the functional group hectare). While two statistically significant structure remained largely unaltered, but relationships were identified (ants & bugs - community composition changed substantially. negative, beetles & spiders - negative), these were This suggests that ant communities are resilient to only apparent on unburnt plots. These results fire-related disturbances (although individual suggest that, at least for these 2 pairs of taxa, and species may be fire sensitive). A similar pattern at the 1ha sampling scale, it may be possible and was evident in the bugs, with a shift within the reliable to use one group as an indicator of the phytophagous (plant-eating) guild from the biodiversity of the other. The fact that disturbance Coccidae, Fulgoridae and Homoptera to the changes the nature of these relationships casts Cicadellidae; and within the predacious guild doubt however on the value of this in studies of from Nabidae to Reduvidae. With the flies impact assessment. however there was a marked reduction (44%) in It was therefore considered that a second fungal feeders (Sciaridae, Drosophilidae, approach, a study of the composition and Mycetophilidae and Scatopsidae) and a substantial structure of these communities, may prove more increase (140%) in phytophagous species useful. If species, or groups of species, could be (primarily Cecidomyiidae), resulting in a found that typified the response of whole taxa, considerable change in community structure. The then these may serve as useful “bio-indicators”. beetles were similarly affected with a 31% An assessment of the effects of disturbance on reduction in the richness of the fungal-feeding ecosystem function would then be feasible, guild (primarily Leionidae), and a 250% increase providing a valuable tool for monitoring progress in richness of phytophagous species (primarily towards, and compliance with, ecologically Chrysomelidae and Curculionidae). The loss of

255 Australia’s Biodiveristy - Responses to Fire

species associated with the decomposer cycle given that they cannot be unambiguously defined implies the frequent burning may be impacting (Caughley and Gunn 1996). It is recognised that a upon nutrient cycling and transfer within these modern conservation strategy cannot however be forests. based solely on areas managed purely for While grouping species into guilds or conservation purposes. Forests (and other functional groups does achieve its intended aim of environments) outside the reserve system will simplifying complex systems, it is at the expense of continue to play an important and complementary considerable important detail. Due to the richness role in meeting conservation objectives with of the communities involved, and the associated respect to biodiversity, even though these forests taxonomic difficulties, this research adopted a may be available for the production of timber and “morphospecies” approach where the biology and other commercial uses in an ecologically ecology of individual species was not examined. sustainable way. York (1996) has shown that, for ants, a knowledge This research has indicated that, through the of the ecology of the constituent species at the development of appropriate management level of genus enables considerable insight into the strategies, we have the knowledge to ensure that mechanisms of change following disturbance. biodiversity is adequately conserved. Strategies Given the aims of this study, the morphospecies required to conserve invertebrate biodiversity are (RBA) approach has however proved to be fundamentally consistent with those used to successful. The limitations of this approach are protect other groups, although the emphasis is acknowledged, and it is considered that additional likely to be placed on the protection of habitats insight could be gained by a more detailed rather than individual species or assemblages (see ecological investigation of individual taxa. This is Samways 1994; New 1995). With regard to proposed in a series of future studies. hazard-reduction burning, the extensive and The limitations of grouping species into frequent use of this management practice has the taxonomic or ecological units are particularly potential to substantially reduce regional evident in the area of conservation biology where biodiversity. The development of strategies which a “species-by-species” management approach may set aside (unburnt) refuges, maintain a diversity of be required, with the ecological needs of each habitats at various stages in the post-fire species addressed separately in Plans of succession, permit variability in other components Management. The incredible diversity of of the fire regime (season of burn and intensity), terrestrial invertebrates may preclude this strategy and allow connectivity between different habitats in Australian forests, with reserve strategies being (corridors), requires urgent attention. Current developed based primarily on vascular plants and draft strategies in forest zoning (eg. SFNSW vertebrates. Whether a conservation strategy 1995) are currently addressing some of these developed using this approach is appropriate for issues, however it is of some concern that other the conservation of invertebrates is questionable, land management agencies still advocate frequent given the lack of concordance between the “broad-acre” burning as a panacea for hazard richness and composition of these very disparate reduction and protection from wildfire without groups of organisms (see Oliver 1995). Similarly, adequate consideration of important biodiversity the use of “ecosystems” or “communities” as issues. ecological units of conservation is problematic

256 Bushfire and forest invertebrates

5. CONCLUSIONS

Infrequent, periodic forest fires (bushfires) are an substantial changes in the composition of species integral part of the modern physical environment assemblages following frequent disturbance of Australian sclerophyll forests. The inherent however, with a loss of taxa dependent upon a variability in natural fire regimes generally results substantial litter layer and stable moist conditions. in a mosaic of habitats with vegetation at different The overall diversity of frequently burnt areas was stages of floristic and structural post-fire maintained by the addition of species with broad succession, each potentially supporting particular tolerances, or adaptations, to drier and more open animal communities. Changes to the components environments. Shifts in community composition of the fire regime (fire intensity, frequency and were best explained by the changes in the amount season of occurrence), as a consequence of forest of leaf litter and insolation at the ground surface, management practices, have the potential to alter habitat elements shown to be dramatically the composition and structure of natural modified by frequent burning. This suggests that communities. The research reported here deals the extensive application of this management with the impact of frequent low-intensity fire practice could result in a reduction in terrestrial (“hazard-reduction burning”) on the abundance, invertebrate biodiversity at a regional scale (γ- richness, composition and structure of terrestrial diversity), with this decrease potentially as high as invertebrate communities. This group, which 50%. constitutes a major component of the overall It was demonstrated that considerable biodiversity in these forests, plays a substantial additional detail concerning, and insight into, the role in the maintenance of ecosystem processes. nature of these changes could be provided by the The ability of the Australian forest industry to inclusion of fairly general information concerning achieve Ecologically Sustainable Management the habitat and dietary preferences of the groups (ESM) depends therefore on a better under investigation. It was apparent that frequent understanding of the impact of commonly used burning leads to a change in the structure of the management strategies on this important invertebrate community. Within species component of the ecosystem. assemblages there were shifts in feeding strategy, While a number of habitat components were with substantial increases in the proportion of responding to large-scale environmental patterns, phytophagous species for bugs, flies and beetles, frequent burning was shown to be impacting upon and a reduction in fly and beetles groups reliant the amount, structure and spatial distribution of primarily on fungal products. With regard to surface leaf litter, the structure and spatial predator guilds, there was a substantial decrease in heterogeneity of components of the vegetation proportional representation for beetles, primarily understorey, moisture levels in the top-soil and the in relation to the family Staphylinidae. For bugs amount and patchiness of insolation reaching the however, the proportion of predators remained ground. While some aspects of these changes are largely unchanged, however there was a total shift likely to reflect post-fire successional trends, the from the family Nabidae to the family Reduviidae. truncation of successional patterns by frequent fire Within one major group of predators, the spiders, and subsequent reduction in environmental there was a substantial reduction in the number of heterogeneity can be expected to impact upon moist-habitat and leaf litter specialists, and a terrestrial invertebrate communities. dramatic increase in the number of species known Using ants, beetles, flies, spiders & bugs as to prefer dry and open environments, particularly representative groups and potential indicators of from the family Zorariidae. Similar patterns were environmental degradation, this research exhibited by the ants, with changes in functional demonstrated that although overall species group representation in response to habitat richness (α-diversity) may not change with alteration. While the impact of these changes on frequent disturbance, species turnover (β- ecosystem function was beyond the scope of this diversity) does. There is a lack of concordance study, substantial changes in the structure of however between groups in the magnitude and invertebrate assemblages and the loss of species direction of these responses. All groups showed associated with the decomposer cycle implies

257 Australia’s Biodiveristy - Responses to Fire

frequent burning may be impacting upon nutrient The application of Rapid Biodiversity Assessment cycling and transfer within these forests. If this is (RBA) methodology demonstrated that the study the case, it would have serious implications with of the composition and structure of communities regard to the maintenance of ecological is likely to prove more rewarding in this regard. sustainability. The identification of individuals to distinct In New South Wales, State Forests has, as a “morphospecies”, while requiring additional stated objective of its Corporate Plan (1992), that it laboratory time and taxonomic expertise, will achieve ecologically sustainable management facilitated the incorporation of broad-level (ESM) by refining concepts and developing ecological information into the assessment and measurable indicators of ecologically sustainable interpretation of environmental impact. use. To be useful therefore, ESM indicators need to Information currently available at the level of be interpretable, significant, cost efficient, and need Family or Genus, but unavailable at the level of to account for variability in space and time, and be species, was sufficient to enable a meaningful appropriate for the scale of management (Turner interpretation of data in relation to impacts on 1993, York 1993). The research reported here community structure and ecological function. supports previous findings concerning the high This in turn enabled the development of spatial variability of invertebrate population management recommendations consistent with numbers, and confirmed the limited use of data the conservation of biological diversity. obtained using coarse-scale taxonomic classification In studies such as this, the accurate (eg. Family or Order). The cost-effectiveness of assessment of the impact of long-term frequent using abundance data alone was shown to be low, disturbance is potentially confounded by short- with high spatial variability and spatial patterning term responses to the most recent perturbation. It requiring large sample sizes to detect management is likely that the rate of post-fire successional effects. This research also identified substantial change will be greater on the frequently disturbed limitations with regard to the use of a single index, plots compared to the unburnt (control) plots, and species richness, as an measure of change and/or the results obtained (and hence assessment of environmental impact. Species richness (α- impact) will be, to some extent, dependent upon diversity) is frequently used to describe and the current successional stage (see Huhta 1971; compare communities, however in this case it was Merrett 1976; York 1994, 1996). In this study the found to provide a deceptive summary of disturbed ares were sampled 2 years after fire and community characteristics. The lack of at a time when there was sufficient fuel to support concordance of richness and abundance patterns another low-intensity fire. With regard to habitat within and between taxa, and the identified modification therefore, this study represents relationships between estimates of richness (α- (potentially) a “worse-case-scenario” typifying one diversity) and turnover (β-diversity) and the spatial end of the spectrum of responses to frequent fire. scale of measurement, meant that the use of these Nevertheless it typifies many areas of dry indices (alone) for impact assessment will sclerophyll forest which are regularly burnt for substantially restrict the level of interpretation that the purposes of hazard reduction. Of more can be derived from the data. Variable spatial concern is the fact that the experimental design patterns in these indices between disturbed and used here substantially “down-plays” the effect of undisturbed sites, and between taxa, also limits recolonisation ability, with potential refuge their applicability in impact assessment. (unburnt) areas no more than 20m from It is often postulated that one group of frequently burnt sites. In a managed forest invertebrates may act as an “indicator” or environment where fuel-reduction burning is “umbrella” group for others, thereby allowing often spatially extensive, the “habitat inferences to be made on the impact of fragmentation” effect is likely to be more disturbance regimes. In this project, it was shown pronounced. Given that many invertebrate species the lack of correlation between taxa with regard to within a community are of low abundance (see richness indices (α- and β-diversity) restricted York 1994) or are habitat or dietary specialists their utility in this regard, primarily as a (York 1996), the risk of local extinction is high. consequence of the non-concordant spatial While local extinction of invertebrates is likely to patterning of these community descriptors and be a regular occurrence in natural systems associated implications for sampling effectiveness. (Samways 1994), systems prone to

258 Bushfire and forest invertebrates

anthropomorphic disturbance require the strategies, from fire exclusion to frequent burning, establishment of adequate measures for the in-situ can be applied to an area in response to protection of successional stages and their management and conservation needs (see Ridley constituent invertebrate fauna (eg. refuges & 1993). Given that the various successional states reserves) and the establishment of links (ie can provide optimal habitats for certain species corridors) to facilitate recolonisation. This need to (York 1994), and that disturbance is an intrinsic provide undisturbed and secure refuges for species and necessary feature in most natural systems with specialist requirements and limited dispersal (Pickett and White 1985), broad-scale fire abilities is the same dilemma facing those exclusion is not a practical management and concerned with the conservation of flowering conservation option. With the increasing plants and vertebrates (see Caughley and Gunn awareness however of the importance of 1996), the difference however is that the groups maintaining environmental and biological involved may be responding to different diversity at a range of spatial scales, the environmental cues and strategies developed for development and implementation of appropriate one taxon may not adequately conserve others (see fire “regimes” which take into consideration the Oliver 1995). scale (ie. frequency and extent) of disturbance, is The development of a “Forest Zoning within the grasp of forest management agencies. System” in New South Wales (SFNSW 1995) should be seen as a necessary response to this situation, whereby a variety of management

259 Bushfire and forest invertebrates

REFERENCES

Abbott, I., Van Heurck, P. and Wong, L. (1984) Bradstock, R.A. and Myerscough, P.J. (1981) Fire Responses to long-term fire exclusion : effects on seed release and the emergence physical, chemical and faunal features of and establishment of seedlings in Banksia litter and soil in a Western Australian forest. ericifolia L.f. Aust. J. Bot. 29:521-532. Aust. For. 47:237-242. Brian, M.V., Mountford, M.D., Abbot, A. and Andersen, A.N. (1985) Seed predation by insects in Vincent, S. (1976) The changes in ant sclerophyllous vegetation at Wilson’s Promontory, species distribution during ten years post- Victoria, with particular reference to ants. fire regeneration of a heath. J. Anim. Ecol. Unpubl. PhD Thesis, School of Botany, 45:115-133. University of Melbourne. 241 pp. Burbidge, A.H., Leicester, K., McDavitt, S. and Andersen, A.N. (1988) Immediate and longer- Majer, J.D. (1992) Ants as indicators of term effects of fire on seed predation by ants disturbance at Yanchep National Park, in sclerophyllous vegetation in south-eastern Western Australia. J.Royal Soc.West.Aust. Australia. Aust. J. Ecol. 13:285-293. 75:89-95. Andersen, A.N. (1990) The use of ant Campbell, A.J. (1973) The Effects of Prescribed communities to evaluate change in Burning on Surface Active Invertebrate Fauna Australian terrestrial ecosystems: a review in Pine and Eucalypt Forest Within the and a recipe. Proc. Ecol. Soc. Aust. 16, 347- Australian Capital Territory. BSc. (Hons) 357. Thesis. ANU. Canberra, ACT. Australia. Andersen, A.N. and Ashton, D.H. (1985) Rates of Campbell, A.J. and Tanton, M.T. (1981) Effect of seed removal by ants at heath and woodland fire on the invertebrate fauna of soil and sites in southeastern Australia. Aust.J.Ecol. litter of a eucalypt forest. In: Fire and the 10:381-390. Australian Biota. (Eds Gill, A.M., Groves, Andersen, A.N. and McKaige, M.E. (1987) Ant R.H. and Noble, I.R.). Australian Academy communities at Rotamah Island Victoria, of Science, Canberra. with particular reference to disturbance and Cary, G.J. and Morrison, D.A. (1995) Effects of Rhytidoponera tasmaniensis. Proc. Roy. Soc. Vict. fire frequency on plant species composition 99: 141-146. of sandstone communities in the Sydney Beattie, A.J. (1995) (ed.) Biodiversity, Australia’s region: Combinations of inter-fire intervals. Living Wealth. Research Unit for Aust.J.Ecol. 20:418-426. Biodiversity and Bioresources, School of Catling, PC. (1991) Ecological effects of Biological Sciences, Macquarie University, prescribed burning practices on the Australia. 131pp. mammals of southeastern Australia. In: Beattie, A.J. and Oliver, I. (1994) Taxonomic Conservation of Australia’s Forest Fauna. (ed minimilisation. Trends in Ecology and Lunney, D.), pp. 353-63. Royal Zoological Evolution 9:488-490. Society of NSW, Mosman. Benson, D.H. (1985) Maturation periods for fire Caughley, G. and Gunn, A. (1996) Conservation sensitive shrub species in Hawkesbury Biology in Theory and Practice. Blackwell sandstone vegetation. Cunninghamia 1:339- Science, Massachusetts USA. 459pp. 349. Christensen, N. (1985) Shrubland fire regimes Birk, E.M. and Bridges, R.G. (1989) Recurrent and their evolutionary consequences. Pp86- fires and fuel accumulation in even-aged 99 in S.T.A. Pickett and P.S.White (Eds) The blackbutt (Eucalyptus pilularis) forests. For. Ecology of Natural Disturbance and Patch Ecol. and Manage. 29:59-79. Dynamics. Academic Press, Orlando, Florida. Bornemissza, G.F. (1969) The re-invasion of burnt woodland areas by insects and mites. Proc. Ecol. Soc. Aust. 4:138.

261 Australia’s Biodiveristy - Responses to Fire

Commonwealth of Australia (1983) National Forestry Commission of NSW (1992) 1992-1995 Conservation Strategy for Australia. Living Corporate Plan. For.Com.NSW, Sydney. Resource Conservation for Sustainable 26pp. Development. Department of Home Affairs Fox, B.J. (1979) An objective method of and Environment. Aust.Govt.Pub.Service, measuring the vegetation structure of animal Canberra. habitats. Aust.Wildl.Res. 6:297-303. CONCOM (1989) Council of Nature Fox, M.D. and Fox, B.J. (1982) Evidence for Conservation Ministers - Australian interspecific competition influencing ant statement on invertebrates. Myrmecia 25, species diversity in regenerating heathland. 123-125. pp.99-110 in Buckley, R.C. (1982) (ed.) Ant- Cornaby, B.W. (1977) Saprophagous organisms Plant Interactions in Australia. Junk and problems in applied resource Publishers. The Hague. partitioning. pp. 96-100 in Mattson, W.J. Fox, M.D. and Fox, B.J. (1986) The effect of fire (ed) (1977) The Role of Arthropods in Forest frequency on the structure and floristic Ecosystems. Springer-Verlag, New York. composition of a woodland understorey. Coy, R. (1996) The impact of fire on soil Aust. J. Ecol. 11:77-85. invertebrates in E.regnans forest at Friend, G. (1996) Fire ecology of invertebrates - Powelltown, Victoria. In: Fire and implications for nature conservation, fire Biodiversity. The Effects and Effectiveness of Fire management and future research. In Fire and Management. pp. 183-198. Biodiversity Biodiversity. The Effects and Effectiveness of Fire Series, Paper No. 8. Biodiversity Unit, Management. Pp. 155-162. Biodiversity Department of the Environment Sport and Series, Paper No. 8. Biodiversity Unit, Territories. Canberra, Australia. Department of the Environment Sport and CSIRO (1991) The Insects of Australia. Second Territories. Canberra, Australia. Edition. Melbourne University Press. Friend, J.A. and Richardson, A.M.M. (1986) Dindal, D.L. and Metz, L.J. (1977) Community Biology of terrestrial amphipods. Ann. Rev. structure of Collembola affected by fire Entomol. 31: 25-48. frequency. pp.88-95 in The Role of Arthropods Gill, A.M. (1981) Adaptive responses of Australian in Forest Ecosystems. Mattson, W.J. (ed) vascular plant species to fires. In: Fire and the Springer-Verlag, New York. Australian Biota (eds. A.M. Gill, R.H. Groves Dolva, G. (1993) The effect of fire on the ecology and and I.R. Noble), pp.243-272. Aust. Acad. life-history of the wood cricket Nambungia Sci., Canberra. balyarta (Nemobiinae:Gryllidae). Unpublished Goldstein, E.L. (1975) Island biogeography of Masters Thesis, University of Western ants. Evolution 29:750-762. Australia. Greenslade, P. and Greenslade, P.J.M. (1987) Duncan, K.W. (1969) The ecology of two species Ecological strategies in collembola: a new of terrestrial Amphipoda (Crustaceae: approach to the use of terrestrial Family Tallitridae) living in waste grassland. invertebrates in environmental assessment. Pedobiologia 9:323-341. In: B.R. Striganova (Ed.), Soil Fauna and Soil Elmes, G.W. and Wardlaw, J.C. (1982) A fertility, pp.245-252. Nauka, Moscow. population study of the ants Myrmica Greenslade, P.J.M. (1984 ) Invertebrates and sabuletti and Myrmica scabrinodis, living at two environmental assessment. Environment and sites in the south of England. II. Effect of Planning 3:13-15. above-nest vegetation. J.Anim.Ecol. 51:665- Greenslade, P.J.M. (1985) Pterygote insects and 680. the soil: their diversity, their effects on soils Engelmann, M.D. (1961) The role of soil and the problem of species identification. arthropods in the energetics of an old field Quaest. Ent. 21:571-585. community. Ecol. Monogr. 31:221-238. ESDWG-Forest Use (1991) Ecologically Sustainable Development Working Groups. Final Report - Forest Use. Australian Government Publishing Service, Canberra. 227pp.

262 Bushfire and forest invertebrates

Greenslade, P.J.M. and Mott, J.J. (1979) Ants of Kremen, C. (1992) Assessing the indicator native and sown pastures in the Katherine properties of species assemblages for natural area, Northern Territory, Australia. areas monitoring. Ecological Applications (Hymenoptera:Formicidae). Pp.153-156 in 2:203-217. Crosby, T.K. and Pottinger, R.P. (eds) Lambert, M.J. (1982) Methods for Chemical proceedings of The 2nd Australasian Analysis. Forestry Commission of NSW. Conference on Grassland Invertebrate Ecology. Sydney, Australia. 3rd Edition. Palmerston North, New Zealand. 22-26th Lawrence, J.F. and Britton, E.B. (1994) Australian May, 1978. Wellington Govt. Printer. Beetles. Melbourne University Press. 192pp. Greenslade, P.J.M. and Thompson, C.H. (1981) Leonard, B.V. (1972) The effect of fire upon selected Ant distribution, vegetation and soil small mammals and leaf litter fauna in relationships in the Coolola-Noosa River sclerophyll forest in southern Australia. Unpubl. area, Queensland. In: Vegetation Classification MSc. Thesis. Monash University, Victoria, in Australia (eds A.N. Gillson & D.J. Australia. Anderson), pp. 192-207. CSIRO and Lowman, M.D. (1982) Seasonal variation in insect Australian National University Press, abundance among three Australian rain Canberra. forests, with particular reference to Harvey, M.S. and Yen, A.L. (1989) Worms to phytophagous types. Aust. J. Ecol. 7:353-361. Wasps. An Illustrated Guide to Australia’s MacArthur, R.H. and MacArthur, J.W. (1961) On Terrestrial Invertebrates. Oxford University bird species diversity. Ecology 42:594-598. Press. 203pp. MacFadyen, A. (1962) Soil arthropod sampling. Hobbs, R.J. (1992) Function of biodiversity in Adv. Ecol. Res. 1:1-34. Mediterranean ecosystems in Australia: MacQuillan, P.B. (1985) The identification of definitions and background. In: Biodiversity of root-feeding cockchafer larvae (Coleoptera: Mediterranean Ecosystems in Australia (ed. R.J. Scarabaeidae) found in pastures in Tasmania. Hobbs), pp. 1-25. Surrey Beatty & Sons. Aust.J.Zool. 33: 509-46. Holldobler, B. and Wilson, E.O. (1990) The Ants. Majer, J.D. (1980) Report on a study of Springer-Verlag, Berlin. 732pp. invertebrates in relation to the Kojonup Huhta, V. (1971) Succession in the spider Nature Reserve fire management plan. Bull. communities of the forest floor after clear- Dept. Biol. West. Aust. Inst. Tech. 2:1-22. cutting and prescribed burning. Ann. Zool. Majer, J.D. (1984) Ant return in rehabilitation Fennici. 8:483-542. mines - an indication of ecosystem resilience. Huhta, V.E., Karpinnen, E., Nurminen, M. and In : B.Dell (ed.), Medicos IV:Proceedings 4th Valpus, A. (1967) Effects of silvicultural International Conference on Mediterranean practices upon arthropod annelid and Ecosystems, pp.105-106, Botany Dept., nematode populations in coniferous forest University of Western Australia, Nedlands. soil. Anneles Zoologie Fennici 4:87-135. Majer, J.D. (1985) Recolonisation by ants of Hurlbert, S.H. (1984) Pseudoreplication and the rehabilitated mineral sand mines on north design of ecological field experiments. Ecol. Stradbroke Island, Queensland, with Monogr. 54:187-211. particular reference to seed removal. Aust. J. IUCN (1980) International Union for the Ecol. 10:31-48. Conservation of Nature, United Nations Majer, J.D. (1992) Ant recolonisation of Environment Program (UNEP) and World rehabilitated bauxite mines of Pocos de Wildlife Fund (WWF) - World Caldas, Brazil. J.Trop.Ecol. 8:97-108. Conservation Strategy - Living Resource Majer, J.D., Satori, M., Stone, R. and Perriman, Conservation for Sustainable Development. W.S. (1982) Recolonisation by ants and IUCN, Gland, Switzerland. other invertebrates in rehabilitated mineral Krebs, C.J. (1985) Ecology. The Experimental sand mines near Eneabba, Western Australia. Analysis of Distribution and Abundance. Third Reclaim. Reveg. Res. 1:63-81. Edition. Harper & Row, New York. 800pp. Majer, J.D., Day, J.E., Kabay, E.D. and Perriman, W.S. (1984) Recolonisation by ants in bauxite mines rehabilitated by a number of different methods. J. Appl. Ecol. 21:355-375.

263 Australia’s Biodiveristy - Responses to Fire

Margules, C.R., Nicholls, A.O. and Pressey, R.L. NFPS (1992) National Forest Policy Statement. A (1988) Selecting networks of reserves to New Focus for Australia’s Forests. maximise biological diversity. Biological Commonwealth of Australia. Canberra. Conservation 43:63-76. Nichols, O.G. and Bunn, S. (1980) Termite Mattson, W.J. (ed) (1977) The Role of Arthropods in utilisation of rehabilitated bauxite mined Forest Ecosystems. Springer-Verlag, New York. areas. Alcoa of Australia Limited. Environ. Mawson, P.R. (1986) A comparative study of Res. Bull.No.9. arachnid communities in rehabilitated Nieuwenhuis, A. (1987) The effect of fire bauxite mines. W.A.I.T. School of Biology frequency on the sclerophyll vegetation of Bulletin No.14. the West Head, New South Wales. McIver, J.D., Parsons, G.L. and Moldenke, A.R. Aust.J.Ecol. 12:373-385. (1992) Litter spider succession after clear- Noble, I.R. and Slatyer, R.O. (1981) Concepts and cutting in a western coniferous forest. models of succession in vascular plant Canad.J.For.Res. 22:984-992. communities subject to recurrent fire. In Fire Merrett, P. (1976) Changes in the ground-living and the Australian Biota (Gill, A.M., Groves, spider fauna after heathland fires in Dorset. R.H. and Noble, I.R. eds.), pp.311-315. Bull. Br. Arachnol. Soc. 3:214-221. Aust.Acad.Sci., Canberra. Metz, L.J. and Dindal, D.L. (1975) Collembola NSESD (1992) National Strategy for Ecologically populations and prescribed burning. Environ. Sustainable Development. Commonwealth Entomol. 4:583-587. of Australia. 128pp. Metz, L.J. and Farrier, M.H. (1973) Prescribed Oliver, I. (1995) Rapid Biodiversity Assessment and burning and populations of soil mesofauna. its Application to Invertebrate Conservation in Environ. Entomol. 2:433-440. Production Forests. Unpubl. PhD Thesis. Morris, M.G. (1975) Preliminary observations on School of Biological Sciences, Macquarie the effects of burning on the Hemiptera University, Australia. (Homoptera and Auchenorhyncha) of Oliver, I. and Beattie, A. J. (1993) A possible limestone grassland. Biol. Conserv. 7:312-319. method for the rapid assessment of Moulton, T.P. (1982) The effect of prescribed burning biodiversity. Cons.Biol. 7:562-568. and simulated burning on soil and litter Oliver, I. and Beattie, A. J. (1996a) Invertebrate arthropods in open forest at Cordeaux, N.S.W. morphospecies as surrogates for species: a Australia. Unpubl. PhD. Thesis. Macquarie case study. Cons. Biol. 10:99–109. University. Oliver, I. and Beattie, A. J. (1996b) Designing a Neumann, F.G. (1992) Effects of low intensity cost-effective invertebrate survey: a test of prescribed fire on invertebrates in mixed methods for rapid biodiversity assessment. eucalypt forest - Wombat State Forest. In Ecological Applications 6:594-607. Tolhurst, K. And Flinn, D. (eds.) Ecological Palmer, M.W. (1993) Putting things in even better Impacts of Fuel Reduction Burning in Dry order: The advantages of canonical Sclerophyll Forest: First Progress Report. correspondence analysis. Ecology 74:2215- Conservation and Environment, Victoria, 2230. Australia. Pickett, S. T. A. and White, P. S. (1985) (Eds) The Neumann, F.G. and Tolhurst, K. (1991) Effects of Ecology of Natural Disturbance and Patch fuel reduction burning on epigeal arthropods Dynamics. Academic Press, Orlando, Florida. and earthworms in dry sclerophyll eucalypt Punttila, P., Haila, Y., Payunen, T and Tukia H. forest of west-central Victoria. Aust. J. Ecol. (1991) Colonisation of clearcut forests by 16, 315-330. ants in the southern Finnish taiga: a New, T.R. (1984) Insect Conservation: An Australian quantitative survey. Oikos 61, 250-262. Perspective. Dr. W.Junk Publishers, Ridley, D. (1993) Fire fuel management in Eden Netherlands. State Forest District. Pp. 91-101 in Ross, J. New, T.R. (1995) Introduction to Invertebrate (Ed) The Burning Question: Fire Management Conservation Biology. Oxford University in NSW. Proceedings of a Conference held Press, New York. 194pp. at the Department of Continuing Education, University of New England, Armidale.

264 Bushfire and forest invertebrates

Samways, M. J. (1994) Insect Conservation Biology. WCED (1987) World Commission on Chapman & Hall, London. 358 pp. Environment and Development. Our Seastedt, T. R. (1984) Below-ground Common Future. Oxford University Press, macroarthropods of annually burned and Oxford, UK. 400pp. unburned tallgrass prairie. Am. Midl. Nat. Weir, E.G. (1978) Development of a Fire Policy. 111:405-408. Parks and Wildlife, Vol.2. National Parks Springett, J. A. (1976) The effect of prescribed and Wildlife Service of N.S.W. burning on the soil fauna and litter Welch, R.C. (1978) Changes in the distribution of decomposition in Western Australian forests. the nests of Formica rufa L. Aust.J.Ecol. 1:77-82. (Hymenoptera:Formicidae) at Brian Woods Springett, J. A. (1979) The effects of a single hot National Nature Reserve, Kent, during the summer fire on soil fauna and on litter decade following coppicing. Insectes Sociaux decomposition in jarrah (Eucalyptus 25:173-186. marginata) forest in Western Australia. Aust. Whelan, R.J., Langedyk, W. and Pashby, A.S. J. Ecol. 4:279-91. (1980) The effects of wildfire on arthropod SFNSW (1995) Forest Management Zoning in State populations in a Jarrah - Banksia woodland. Forests (Draft). State Forests of New South W.A. Naturalist 14:214-220. Wales. Sydney, Australia. Whelan, R.J. (1995) The Ecology of Fire. Taylor, R. W. (1983) Systematics and ecology: Cambridge University Press. 346pp. reflections from the interface. Pp50-79 in Williams, J.E. and Gill, A.M. (1995) The Impact of E.Highley and R.W. Taylor (eds) Australian Fire Regimes on Native Forests in Eastern New Systematic Entomology: A Bicentenary South Wales. Environmental Heritage Perspective. CSIRO, Melbourne. Monograph Series No. 2. NSW National Ter Braak, C. J. F. (1986) Canonical Parks and Wildlife Service. 68pp. Correspondence Analysis: a new eigenvector Woodward, F.I. (1993) How many species are technique for multivariate direct gradient required for a functioning ecosystem? In analysis. Ecology 67:1167-1179. Ecosystem Function of Biodiversity. Eds. E.D. Ter Braak, C. J. F. (1991) CANOCO version 3.12. Schultz and H.A. Mooney. Springer-Verlag, Agricultural Mathematics Group, DLO Box Heidelberg. 100, 6700 AC, Wageningen, The Yeatman, E.M. and Greenslade, P.J.M. (1980) Netherlands. Ants as indicators of habitat in three Tolhurst, K. G., Kellas, J. D. and Wilson, A. A. conservation parks in South Australia. (1992) Low intensity fire behaviour and fuel S.Aust.Nat. 55:20-26, 30. dynamics in dry sclerophyll forest, Wombat Yen, A.L. (1987) A preliminary assessment of the State Forest. In Tolhurst, K. And Flinn, D. correlation between plant, vertebrate and (eds.) Ecological Impacts of Fuel Reduction Coleoptera communities in the Victorian Burning in Dry Sclerophyll Forest: First mallee. In : Majer, J.D. (ed.), The Role of Progress Report. Conservation and Invertebrates in Conservation and Biological Environment, Victoria, Australia. Survey pp.73-88. W.A. Dept. of Turner, J. (1993) Research and Ecologically Conservation and Land Management Sustainable Forest Management. A Discussion Report. Paper. Research Division, State Forests of York, A. (1989) The Response of Ant Communities to New South Wales. 26pp. Fire-induced Habitat Disturbance. Van der Drift, J. (1958) The role of the soil fauna Unpublished PhD. Thesis, School of in the decomposition of forest litter. pp.357- Biological Sciences, University of New 360 in Proc. 15th Intern. Congr. Zool. South Wales. Van Loon, A.P. (1970) Further Investigations in York, A. (1993) Towards ecologically sustainable the Effects of Prescribed Burning on Young forest use. In Rapid Biodiversity Assessment (ed Even-aged Blackbutt. Taree Forestry Office Beattie, A.J.), pp. 51-55. Proceedings of the Research Report. Biodiversity Workshop Macquarie University - 1993. Research Unit for Biodiversity and Bioresources, School of Biological Sciences, Macquarie University.

265 Australia’s Biodiveristy - Responses to Fire

York, A. (1994) The long-term effects of fire on forest ant communities: management implications for the conservation of biodiversity. Memoirs of the Queensland Museum 36, 231-239. York, A. (1996) Long-term effects of fuel reduction burning on invertebrates in a dry sclerophyll forest. In Fire and Biodiversity. The Effects and Effectiveness of Fire Management. Pp. 163-181. Biodiversity Series, Paper No. 8. Biodiversity Unit, Department of the Environment Sport and Territories. Canberra, Australia. Zborowski, P. and Storey, R. (1995) A Field Guide to Insects in Australia. Reed Books, Sydney. 207pp. Zedler, P.H., Gautier, C.R. and McMaster, G.S. (1983) Vegetation change in response to extreme events: the effect of a short interval between fires in California chaparral and coastal scrub. Ecology 64:809-818.

266

Minerva Access is the Institutional Repository of The University of Melbourne

Author/s: YORK, A

Title: Long-term effects of repeated prescribed burning on forest invertebrates: management implications for the conservation of biodiversity

Date: 1999

Citation: YORK, A. (1999). Long-term effects of repeated prescribed burning on forest invertebrates: management implications for the conservation of biodiversity. Gill, AM (Ed.). Woinarski, JCZ (Ed.). YORK, A (Ed.). Australia's Biodiversity - Responses to Fire. Plants, birds and invertebrates., (Biodiversity Technical Paper, No. 1), pp.181-266. Department of Environment and Heritage.

Persistent Link: http://hdl.handle.net/11343/127530

File Description: Published version