Population processes in an early successional heathland: a case study of the eastern chestnut mouse (Pseudomys gracilicaudatus)
by
Felicia Pereoglou
A thesis submitted for the degree of Doctor of Philosophy of The Australian National University March 2016
Grarock K., Introduction, spread, impact and control of the common myna
Grarock K., Introduction, spread, impact and control of the common myna
Candidate's Declaration
This thesis contains no material that has been accepted for the award of any other degree or diploma in any university. The research, analysis and writing in the thesis are substantially (>90%) my own work. To the best of my knowledge, it contains no material previously published or written by another person, except where due reference is made in the text.
Felicia Pereoglou Date: 14 March 2016
Pereoglou, F., Population processes in an early successional heathland iii
Acknowledgements
I thank all the people who have dedicated their time, provided motivation, support and advice during the course of this work. I sincerely thank the members of my supervisory panel – Professor David B. Lindenmayer, Dr Sam Banks, Mr Christopher MacGregor, Dr Jeffrey Wood, and Dr Fred Ford – for their guidance and support. In particular, Sam and Chris who without your assistance through those particularly tough times, I would not have completed this work at all. I also extend thanks to;
The volunteers who assisted with fieldwork surveys. A very special thanks to Wendy Hartman who made those many early mornings and nest assessments so much more enjoyable, and to Alfred Wellington who taught me a lot too.
Staff at the Fenner School of Environment and Society for their assistance throughout the life of this project, especially Phil Greaves, Piers Bairstow, Mauro Davanzo, Amy Chen, Diane Jakobasch, and Claire Shepherd
My family, friends and colleagues for their help, advice, motivation and patience
Pereoglou, F., Population processes in an early successional heathland iv
Abstract
The eastern chestnut mouse (Pseudomys gracilicaudatus) has been described as a fire specialist inhabitant of early successional heathland. There is a global concern that changes to natural disturbance regimes will place early successional habitat specialists like the eastern chestnut mouse at an increased risk of extinction by altering landscape patterns of habitat suitability. Despite this concern, the fundamental population processes that occur in early successional heathlands are currently unknown for many species. This thesis comprises four research articles that investigate the underlying mechanisms used by a species to persist in this spatially and temporally variable habitat. To determine which features of burned heath were selected as refuge sites, 58 individuals of the eastern chestnut mouse were radio-tracked. Animals preferentially selected multiple aboveground shelter sites characterised by tall, dense vegetation with structural features present. The eastern chestnut mouse was ephemeral and non-gregarious in its refuge use. Refugia were highly flammable. These findings imply land management should ensure retention of vegetation structure on the ground layer and dense habitat patches in burned areas to assist with population viability. A long-term monitoring data set was analysed to evaluate the effects of landscape structure, fire history, and site-level habitat quality on site occupancy in an early successional heathland. Eastern chestnut mouse site occupancy was strongly related to patch area and landscape heterogeneity. There was no direct effect of recent or historical fire, but site occupancy was associated with dead shrubs in the understory and rock cover on ground layer. Small mammal species associations did not conform to that predicted by contemporary ecological theory. The findings suggest landscape configuration and continuity be used to guide survey work and fire regimes to promote eastern chestnut mouse population growth should retain critical habitat features rather than be based on temporal rates of successional stages. To examine population genetic structure, dispersal patterns and the effects of landscape variation on dispersal of an early successional specialist, microsatellite markers were used to genotype tissue obtained from 287 individuals. Differentiation between sites fitted an isolation- by-distance pattern. There was spatial autocorrelation of multilocus genotypes to a distance of 2–3 km with no evidence for sex-biased dispersal. Recently burned vegetation had greater conductance for gene flow than unburnt habitat, but variation in habitat quality between vegetation types and occupied patches had no effect on gene flow. These findings indicate a ‘generalist’ dispersal strategy is advantageous for post-fire specialists to access early successional vegetation. To quantify survival, recruitment and immigration rates in populations of the eastern chestnut mouse, demographic modelling and genetic assignment tests were applied to two years of mark–recapture data. Survival and recruitment rates varied across sites. Survival was Pereoglou, F., Population processes in an early successional heathland v
influenced by fire history and elements of habitat structure affected by fire. Immigration contributed very little to recruitment compared to local reproduction, suggesting populations are demographically discrete at this post-fire stage. Demographic and genetic data were consistent; sites with high recruitment also had higher genetic diversity and lower temporal genetic structure, consistent with reduced genetic drift. This thesis provides an investigation of the strategies used by the eastern chestnut mouse to persist in an early successional heathland and adds significant new knowledge on the critical resources, predictors of occurrence, population genetics, and demographic rates of the species. The major implication of this research is that a predicted increase in fire occurrence will potentially influence the viability of populations of the eastern chestnut mouse by promoting gene flow but concomitantly decreasing critical habitat features that support site occupancy, increase survival rates, and provide nesting refugia. Furthermore, landscape continuity has a secondary influence population processes and this must be considered in conjunction with the disturbance regime to inform management actions.
Pereoglou, F., Population processes in an early successional heathland vi
Table of Contents
Candidate's Declaration ...... iii
Acknowledgements ...... iv
Abstract ...... v
Table of Contents ...... vii List of Figures ...... xi List of Tables ...... xii
Chapter 1: Introduction ...... 13 1.1 Introduction ...... 14 1.1.1 Early successional ecosystems ...... 14 1.1.2 Eastern chestnut mouse ...... 15 1.1.2.1 Origin, taxonomy and conservation status ...... 15 1.1.2.2 Distribution ...... 15 1.1.2.3 Response to fire ...... 15 1.1.2.4 Habitat ...... 16 1.1.2.5 Appearance ...... 16 1.1.2.6 Diet ...... 17 1.1.2.7 Breeding, growth and longevity ...... 17 1.2 Context and structure ...... 18 1.2.1 Chapter 2 – Refuge site selection in recently burned heath ...... 18 1.2.2 Chapter 3 – Early successional heathland site occupancy ...... 19 1.2.3 Chapter 4 – Landscape genetics in disturbance-prone environments ...... 19 1.2.4 Chapter 5 – Estimating survival and recruitment ...... 20 1.2.5 Chapter 6 – Conclusion ...... 20 1.3 Thesis summary ...... 20 1.4 Publication summary, candidate contribution and co-author acknowledgement ... 21 1.5 References ...... 23
Chapter 2: Refuge site selection ...... 26 2.1 Abstract ...... 27 2.2 Introduction ...... 27 2.3 Materials and methods ...... 29 2.3.1 Study sites ...... 29 2.3.2 Trapping and radio-collaring ...... 30 2.3.3 Assessment of shelter sites ...... 31 2.3.4 Data analysis ...... 31 2.4 Results ...... 32
Pereoglou, F., Population processes in an early successional heathland vii
2.4.1 Trapping and radio-tracking ...... 32 2.4.2 Shelter site selection ...... 32 2.4.3 Refuge use and fidelity ...... 34 2.5 Discussion ...... 35 2.5.1 Refuge site selection ...... 35 2.5.2 Refuge site use and fidelity ...... 36 2.5.3 Prescribed burning practices ...... 37 2.5.4 Conclusion ...... 38 2.6 Acknowledgements ...... 39 2.7 References ...... 39
Chapter 3: Site occupancy ...... 45 3.1 Abstract ...... 46 3.2 Introduction ...... 46 3.3 Materials and Methods ...... 49 3.3.1 Study species ...... 49 3.3.2 Study area and survey design ...... 50 3.3.3 Statistical analysis ...... 53 3.4 Results ...... 54 3.4.1 Correlation between explanatory variables ...... 54 3.4.2 Fire history ...... 55 3.4.3 Site-level habitat quality ...... 55 3.4.4 Species associations...... 55 3.4.5 Landscape structure...... 58 3.4.6 ‘Best’ model for predicting eastern chestnut mouse site occupancy ...... 59 3.5 Discussion ...... 59 3.5.1 Importance of landscape structure versus site-level habitat selection ...... 59 3.5.2 Fire history as a predictor of site occupancy ...... 61 3.5.3 Species interactions and the habitat accommodation model ...... 62 3.5.4 Conclusion ...... 62 3.6 Acknowledgement ...... 63 3.7 References ...... 63
Chapter 4: Landscape genetics ...... 70 4.1 Abstract ...... 71 4.2 Introduction ...... 71 4.3 Methods ...... 73 4.3.1 Study species ...... 73 4.3.2 Study area and sampling ...... 74 4.3.3 DNA extraction, PCR amplification and genotyping ...... 74
Pereoglou, F., Population processes in an early successional heathland viii
4.3.4 Genetic data analysis ...... 75 4.3.4.1 Population genetic summary statistics ...... 75 4.3.4.2 Population subdivision and isolation by distance (IBD) ...... 76 4.3.4.3 Sex-biased dispersal ...... 77 4.3.4.4 Landscape resistance ...... 77 4.4 Results ...... 81 4.4.1 Population genetic summary statistics ...... 81 4.4.2 Population subdivision and spatial genetic structure ...... 81 4.4.3 Sex-biased dispersal ...... 83 4.4.4 Landscape resistance ...... 85 4.5 Discussion ...... 85 4.5.1 Landscape connectivity and gene flow ...... 86 4.5.2 – Population genetic structure and dispersal behaviour ...... 89 4.5.3 – Conclusion ...... 90 4.6 Acknowledgement ...... 90 4.7 References ...... 91
Chapter 5: Population dynamics ...... 97 5.1 Abstract ...... 98 5.2 Introduction ...... 98 5.3 Materials and methods ...... 100 5.3.1 Study species ...... 100 5.3.2 Study area ...... 101 5.3.3 Animal capture and marking ...... 101 5.3.4 Genetic data analysis ...... 102 5.3.5 Estimating survival and recruitment ...... 102 5.4 Results ...... 104 5.4.1 Population genetics summary statistics ...... 104 5.4.2 Population subdivision and spatial genetic structure ...... 104 5.4.3 Assignment tests ...... 104 5.4.4 Survival (CJS) models ...... 106 5.4.5 Recruitment (Pradel) models ...... 109 5.5 Discussion ...... 109 5.5.1 Internal population processes ...... 111 5.5.2 External population processes ...... 112 5.5.3 Conclusion ...... 113 5.6 Acknowledgement ...... 113 5.7 References ...... 113
Chapter 6: Conclusion ...... 119
Pereoglou, F., Population processes in an early successional heathland ix
6.1 Population processes of the eastern chestnut mouse ...... 120 6.2 Population persistence in an early successional community ...... 122 6.2.1 The role of fire ...... 123
Consolidated references ...... 124
Appendix 1: Manuscript – Refuge site selection ...... 143
Appendix 2: Manuscript – Site occupancy ...... 153
Appendix 3: Manuscript – Landscape genetics ...... 169
Appendix 4: Supporting information for Chapter 3 ...... 185
Appendix 5: Supporting information for Chapter 4 ...... 193
Pereoglou, F., Population processes in an early successional heathland x
List of Figures Figure 1.1: The eastern chestnut mouse (Pseudomys gracilicaudatus)...... 17 Figure 2.1: Location of study sites in Booderee and Jervis Bay national parks ...... 30 Figure 2.2: Light intensity measured in lux at 20 cm intervals of a 1m profile through shelter sites and paired control sites...... 33 Figure 3.1: Conceptual diagram showing the predicted effects of explanatory variables on site occupancy...... 48 Figure 3.2: Map of Booderee national park indicating study sites and vegetation types...... 52 Figure 3.3: Layout of a site transect and the location of vegetation plots...... 52 Figure 3.4: Conceptual diagram showing the modelled effects of explanatory variables on site occupancy...... 57 Figure 3.5: Number of eastern chestnut mouse captures predicted from hierarchical generalised linear models over a range of values of significant explanatory variables...... 58 Figure 4.1: Study area and distribution of survey sites ...... 75 Figure 4.2: Habitat quality relating to movement across a null landscape (A), burned areas (B), selected vegetation types (C), and occupied habitat (D) ...... 80 Figure 4.3: Relationship between genetic distance and geographic distance ...... 83 Figure 4.4: Correlograms indicating spatial autocorrelation of pairwise multilocus genetic distances over 5 km and 250m ...... 84 Figure 5.1: Map of Booderee national park indicating study sites and vegetation types...... 101
Pereoglou, F., Population processes in an early successional heathland xi
List of Tables Table 2.1: Trapping and radio-collaring site summary ...... 31 Table 2.2: Means ( s.e.), t-statistics and significance levels for differences in height and total number of structures present between paired shelter and control sites ...... 33 Table 2.3: Proportion of successes, test statistic and significance level for differences in the types of structures present in shelter and control sites ...... 33 Table 2.4: Multiple regression statistics for a test of the effect of type of structure on standardised light intensity ...... 33 Table 2.5: Parameters for predicting presence of a primary refuge ...... 35 Table 3.1: Summarized description of the eastern chestnut mouse and co-occurring small mammal species...... 51 Table 3.2: Description of explanatory variables used to predict site occupancy ...... 54 Table 3.3: Correlation matrix showing Spearman’s rank correlation coefficients for explanatory variables ...... 56 Table 3.4: Ten best generalized linear models from fitting all possible subsets of explanatory variables ...... 57 Table 3.5: Multivariate HGLM of significant explanatory variables ...... 59 Table 4.1: Raster cell values used to create resistance landscapes ...... 79 Table 4.2: Population genetic parameters ...... 82 Table 4.3: Results of Mantel and partial Mantel tests of landscape resistance on gene flow ...... 86 Table 5.1: Eastern chestnut mouse capture data, population genetics summary statistics, and assignment tests by site ...... 105
Table 5.2: Pairwise site FST for the eastern chestnut mouse ...... 106 Table 5.3: Cormack-Jolly-Seber (CJS) and Pradel model averaged parameter estimates ...... 107 Table 5.4: Model summary describing factors affecting survival ...... 108 Table 5.5: Model summary describing factors affecting recruitment ...... 110
Pereoglou, F., Population processes in an early successional heathland xii
Chapter 1: Introduction- make white
Introduction
Pereoglou, F., Population processes in an early successional heathland 13 Chapter 1: Introduction
1.1 Introduction
1.1.1 Early successional ecosystems
Disturbance plays a fundamental role in the formation and maintenance of ecological communities (Miller et al., 2011, Connell and Slatyer, 1977). Succession refers to the observed changes in an ecological community following a disturbance event (Connell and Slatyer, 1977). Early successional ecosystems are the starting point for post-disturbance community regeneration (Swanson et al., 2011). In Australia, wildfire is one of the dominant natural forces in creating early successional ecosystems. Wildfire disturbances can be described by their intensity, season of occurrence, extent and patchiness, frequency, and type (Whelan, 1995). Combinations of these factors generate a disturbance regime (Gill, 1975), which then determines the extent and composition of the early successional ecosystem Many endemic species are dependent on post-fire conditions for survival (Whelan, 1995, Keith et al., 2002, Swanson et al., 2011). Early successional communities are ephemeral ecosystems confined to the variable length of time between a disturbance, and re-establishment of the later successional stages of community composition (Swanson et al., 2011). Temporal variation in the duration of the early successional stage is influenced by the (i) disturbance type (Rossi et al., 2007, Delmoral and Bliss, 1993, Sousa, 1979), (ii) disturbance regime (Bradstock et al., 2012), (iii) habitat type (rainforest, forest, shrubland, heathland, grassland, etc.) (Keith et al., 2002, Gill and Catling, 2002), and (iv) geographic region (Velle and Vandvik, 2014). Research on the processes occurring in early successional ecosystems links disturbance ecology with habitat fragmentation theory and conservation. Early successional ecosystems occur as habitat patches of varying size, shape and connectedness within the wider landscape, incorporating the importance of landscape continuity. Early successional ecosystems are not communities in ‘recovery’ but ones with significant biological value. Because early successional ecosystems, and the species that occur within them, are limited by duration of seral stage, their distribution, and viability of populations occurring within them, can be threatened by disturbance suppression or changes in the prevailing disturbance regime. This thesis explores themes relating to population processes that occur in an early successional heathland in south-eastern Australia. The eastern chestnut mouse (Pseudomys gracilicaudatus, Gould) is used as a case study to investigate critical resource requirements, site occupancy, landscape genetics, and population dynamics of an early successional species and habitat specialist. This research is timely given a predicted increase in the frequency and severity of disturbance events with global climate change (Flannigan et al., 2009, Steffen et al.,
Pereoglou, F., Population processes in an early successional heathland 14 Chapter 1: Introduction
2009, Williams et al., 2001) will require a comprehensive understanding of the mechanisms underlying population recovery of early successional habitat specialists following disturbance.
1.1.2 Eastern chestnut mouse
1.1.2.1 Origin, taxonomy and conservation status
All Australian rodents belong to the family Muridae (Lee et al., 1981, Strahan, 2008). Native mice of the Pseudomys genus are conilurine rodents, or Australian Old Endemic rodents. This group originated in Australia during the Miocene (Lee et al., 1981) and its closest relatives occur in New Guinea (Breed and Ford, 2007). The Pseudomys are the largest group of Australian rodents (Watts and Aslin, 1981) encompassing a heterogeneous assemblage of 24 species (Breed and Ford, 2007). Many are considered the best examples of fire specialist fauna in Australia (Catling and Newsome, 1981). Native rodents are a group that are poorly conserved in Australia with 55% of species considered either threatened or extinct (Breed and Ford, 2007). Many Pseudomys species are of conservation concern. In New South Wales alone, all but one of the 11 species of Pseudomys that occur in the state are listed on threatened species schedules (Dickman et al., 2000). The eastern chestnut mouse (Pseudomys gracilicaudatus) is one such species, considered Vulnerable to extinction through loss of habitat and long-term fire exclusion and currently listed on Schedule 2 Threatened Species Conservation Act 1995 (NSW).
1.1.2.2 Distribution
The eastern chestnut mouse has a disjunct distribution along the east coast of Queensland and New South Wales, Australia. Existing populations of the species are isolated by substantial distances, often encompassing apparently suitable habitat from which the species is absent (Fox, 2008). As such, the species is now considered rare or extinct in many parts of its range (Breed and Ford, 2007). First recorded in New South Wales in 1971 (Mahoney and Posamentier, 1975), the eastern chestnut mouse was thought to be extinct in southern Australia until 2002 when it was discovered in the Jervis Bay region (Meek and Triggs, 1997). This area now supports the only contemporary population of eastern chestnut mouse south of Gosford, New South Wales and is arguably the most isolated in eastern Australia. The northern limit of the eastern chestnut mouse distribution is Cooktown, Queensland.
1.1.2.3 Response to fire
The eastern chestnut mouse was first described as a ‘fire specialist’ by Fox (1982) in a study of small mammal community recovery following wildfire at a single site in northern New South Wales. The time taken for each species to reach a peak in population abundance was plotted along a continuum, which was then used to determine whether the species was an early-,
Pereoglou, F., Population processes in an early successional heathland 15 Chapter 1: Introduction mid- or late-successional community specialist. The eastern chestnut mouse population reached maximum densities within the first few years post-fire before a decline in abundance and replacement by a population of the swamp rat (Rattus lutreolus) (Fox, 1982). Subsequent studies of the eastern chestnut mouse focused on understanding whether changes in species’ abundance over time were a result of the competitive response of the sympatric species (Catling, 1986, Morris et al., 2000, Higgs and Fox, 1993), or a response to increasing vegetation cover and changes in habitat structure post-fire (Monamy and Fox, 2000, Fox et al., 2003). To date, the ‘fire-mouse story’ indicates disturbance events sustain populations of the eastern chestnut mouse by removing vegetation cover and reducing habitat structure, subsequently easing competitive pressure by lowering the population abundance of competitor species. No studies of the eastern chestnut mouse have compared the species’ population size and demographic rates, before and/or after fire. Nor is there any data on habitat use, or local- and landscape-scale movements before and after fire. Whether the eastern chestnut mouse is an early successional/fire specialist because (i) it exclusively colonises burned heath, or (ii) it survives a fire event in situ and then increases population density post-fire, currently remains unknown.
1.1.2.4 Habitat
The eastern chestnut mouse is considered an early seral stage specialist of regenerating heathland (Fox, 1982). Detailed information on eastern chestnut mouse habitat preferences is lacking through limited research in this area. However, of the few studies that have been done, the species has been associated with dense but patchy vegetation in the upper shrub layer, and woody debris on the ground layer (Haering and Fox, 1995). Both vegetation structural attributes and understorey floristic richness also appear to be important (Fox and Fox, 1981). The density of vegetation cover, or rate of vegetation recovery after fire, is considered to have the greatest influence on eastern chestnut mouse abundance (Monamy and Fox, 2000, Monamy and Fox, 2010).
1.1.2.5 Appearance
The Pseudomys are mainly mouse-sized with common characteristics such as a well haired, bicoloured tail (darker on the upper surface), and females with only two pairs of teats, which distinguishes them from the true rats (Rattus) (Watts and Aslin, 1981). Adults of the eastern chestnut mouse show significant sexual dimorphism (Fox and Kemper, 1982); measuring up to 145 mm in length for males and 138 mm for females, and attaining average weights of 90 g for males and 75 g for females (Fox, 2008). The species is named after its characteristic grizzled chestnut fur above and has a pale, greyish belly (Fox, 2008) (Fig. 1.1). The eastern chestnut mouse can be distinguished from the similar appearance of the swamp rat
Pereoglou, F., Population processes in an early successional heathland 16 Chapter 1: Introduction
(Rattus lutreolus) by a pale ring around the eye and ventrally light coloured feet with long, grey- white hairs on the dorsal surface that extend beyond the claws (Fox, 2008).
Figure 1.1: The eastern chestnut mouse (Pseudomys gracilicaudatus).
1.1.2.6 Diet
The eastern chestnut mouse is a generalist herbivore whose diet consists predominantly of seeds and stems, but also includes fungi, insects, leaves, roots and pollen (Luo et al., 1994). The eastern chestnut mouse demonstrates an opportunistic foraging strategy and is less specialised in food selection than any other species within the Pseudomys genus (Luo et al., 1994). This foraging strategy is considered beneficial for an early successional habitat specialist because the ability to take advantage of all available food resources is a basic requirement for survival in recently burned areas (Luo et al., 1994). A decrease in dietary diversity in the later successional stages of vegetation regeneration following fire has been linked to a decline in the species’ density over time. This, in turn, has been attributed to changes in the availability of food resources and increasing interspecific competition (Luo and Fox, 1994).
1.1.2.7 Breeding, growth and longevity
The eastern chestnut mouse breeding season extends from September to March, but in good years can commence in August (Fox, 2008). Female oestrous cycle is 7 – 8 days with a gestation of approximately 27 days (Fox and Kemper, 1982). Females are capable of producing litters of 1 – 5 offspring (usually three) (Fox, 2008). The species is precocial at birth and shows
Pereoglou, F., Population processes in an early successional heathland 17 Chapter 1: Introduction rapid early development – the lower incisors have already erupted at birth, young are fully furred by six days, have opened their eyes by 11 days, and are weaned by the fourth week (Fox and Kemper, 1982). Females exhibit a post-partum oestrous and multiple litters may be produced during a single breeding season (Fox and Kemper, 1982). Although no individuals have been observed to breed in the season of their birth, females have been observed to breed in two successive years and males can live through a second summer (Fox, 2008). A short gestation, precocial young, rapid postnatal development and early weaning accelerate the reproductive rate which is appropriate for a fire-specialist species that reaches maximum abundance in the early successional stage of vegetation regeneration following fire (Fox and Kemper, 1982).
1.2 Context and structure
The following section provides a brief overview of the chapters in the thesis. The body of work is presented as a compilation of separate papers; either already published as journal articles, or submitted journals for publication. As such, it follows The Australian National University’s annotated Thesis by Compilation Guideline of 10 May 2013, available on the following webpage:
1.2.1 Chapter 2 – Refuge site selection in recently burned heath - Pereoglou, et al. (2011) Wildlife Research, 38(4)
Temporal reduction in shelter is an indirect primary impact of fire. Removal of refugia has implications for shelter site selection and fidelity – factors that have been largely overlooked in studies of Australian rodent fauna. In Chapter 2, diurnal refuge sites were identified through radiotelemetry and their habitat attributes were compared with those of a matched set of control sites. Habitat features and fidelity parameters were used to classify refuge site use. Animals were found to preferentially select multiple aboveground shelter sites characterised by tall, dense vegetation with structural features present. These refugia were highly flammable and do not provide protection from fire. Mice were ephemeral and non- gregarious in their refuge use. There was some evidence for inheritance of refuge sites from a parent, as well as inter-season shelter site fidelity. Knowledge of species’ nest requirements and daytime refuge habitat and use is fundamental for conservation efforts in fire-prone landscapes.
Pereoglou, F., Population processes in an early successional heathland 18 Chapter 1: Introduction 1.2.2 Chapter 3 – Early successional heathland site occupancy - Pereoglou, et al. (2016) Landscape Ecology, 31(2)
Changes to natural disturbance regimes can place early successional habitat specialists at an increased risk of extinction by altering landscape patterns of habitat suitability. In Chapter 3, a series of hypotheses were developed to evaluate the effects of fire history, landscape structure, and site-level habitat quality on site occupancy by the eastern chestnut mouse.
Patch area and landscape heterogeneity were strongly related to site occupancy. Site occupancy was associated with dead shrubs in the understorey and rock cover on the ground layer, but was not directly influenced by recent or historical fire. There was no evidence for predictable species associations in the early successional ecosystem. Fire regimes in areas supporting populations of the eastern chestnut mouse should be sensitive to the retention of critical habitat features rather than be based on temporal rates of successional stages. For management to adequately account for post-disturbance patterns in early successional ecosystems, a species-by-species, multi-scaled approach to research, replicated throughout a species’ range is necessary.
1.2.3 Chapter 4 – Landscape genetics in disturbance-prone environments - Pereoglou, et al. (2013) Molecular Ecology, 22(5)
Species that specialise in disturbed habitats may have considerably different dispersal strategies than those adapted to more stable environments. However, little is known of the dispersal patterns and population structure of such species. In Chapter 4, microsatellite markers were used to assess the effect of landscape variation and recent disturbance history on eastern chestnut mouse dispersal. Significant differentiation between sites that fitted an isolation-by-distance pattern was demonstrated, and spatial autocorrelation of multilocus genotypes occurred up to a distance of 2–3 km. There was no consistent genetic evidence for sex-biased dispersal. The influence of different habitat and fire-specific landscape resistance scenarios on genetic distance between individuals were tested. A significant effect of fire was found indicating recently burned vegetation had greater conductance for gene flow than unburnt habitat, but variation in habitat quality between vegetation types and occupied patches had no effect on gene flow. Post-fire specialists must evolve an effective dispersal ability to move over distances that would ensure access to early successional stage vegetation. Natural disturbance and natural heterogeneity may therefore not influence population genetic structure as negatively as expected.
Pereoglou, F., Population processes in an early successional heathland 19 Chapter 1: Introduction 1.2.4 Chapter 5 – Estimating survival and recruitment - Pereoglou, et al. Ecology and Evolution Submitted
Understanding how demographic parameters of post-fire specialists vary with fire history can provide insights into the population viability and distribution of such species, including their response to landscape management and novel fire regimes. In Chapter 5, demographic and genetic analyses were used to quantify survival, recruitment and immigration rates in populations of the eastern chestnut mouse.
Survival and recruitment rates varied across sites. Survival models that included fire history and elements of habitat structure affected by fire had the most support. Recruitment was attributed to immature animals entering the adult breeding population prior to the breeding season. Recruitment through immigration was negligible compared to the survival and reproduction of local residents. High levels of recruitment corresponded to increased genetic diversity within a site and lower temporal genetic structure, consistent with reduced genetic drift. In light of landscape genetics analyses showing that fire promotes population connectivity, this study suggests a two-phase model of spatial population dynamics, where post- fire colonization is followed by locally driven population dynamics, in which survival rates are influenced by fire-history effects on habitat quality. Thus, post-fire specialists can be dependent on landscape-level connectivity and site-level habitat quality, both of which are influenced by fire history.
1.2.5 Chapter 6 – Conclusion
Chapter 6 discusses the major findings of this thesis: (i) the indirect importance of fire in creating and maintaining habitat features that effect the selection of diurnal refugia, success of site occupancy, and demographic rates of habitat specialists of early successional heathland; and (ii) the life history strategies of species that occur in spatially and temporally variable environments as demonstrated by the eastern chestnut mouse i.e. fire-mediated gene flow, ‘generalist’ dispersal, recruitment through local reproduction, etc.
1.3 Thesis summary
This thesis comprises four research articles that combine to provide a detailed investigation of the population processes occurring in an early successional heathland. This research adds significant new knowledge on the refuge requirements, predictors of site occurrence, landscape genetics, and population dynamics, of the eastern chestnut mouse in a spatially and temporally variable environment. These findings are timely due to the growing concern about managing fire-prone landscapes and attempts to conserve species that specialise
Pereoglou, F., Population processes in an early successional heathland 20 Chapter 1: Introduction in ephemeral ecosystems originating from disturbance events.
1.4 Publication summary, candidate contribution and co- author acknowledgement
The material presented in this thesis is an accurate representation of research conducted during my candidature for the degree of Doctor of Philosophy between February 2008 and March 2016. This material is presented as contributions to the published scientific literature. These contributions are listed below in order of their appearance in this thesis: (i) Pereoglou, F., MacGregor, C., Banks, S.C., Ford, F., Wood, J., Lindenmayer, D.B. (2011) Refuge site selection by the eastern chestnut mouse in recently burned heath. Wildlife Research, 38(4): 290-298. doi:10.1071/WR11007 (Appendix 1). (ii) Pereoglou, F., MacGregor, C., Banks, S.C., Wood, J., Ford, F. Lindenmayer, D.B. (2016) Landscape, fire and habitat: which features of recently burned heathland influence site occupancy of an early successional specialist? Landscape Ecology, 31(2): 255-269. doi:10.1007/s10980-015-0240-2 (Appendix 2). (iii) Pereoglou, F., Lindenmayer, D.B., MacGregor, C., Ford, F., Wood, J., Banks, S.C. (2013) Landscape genetics of an early successional specialist in a disturbance- prone environment. Molecular Ecology 22(5): 1267-1281. doi:10.1111/mec.12172 (Appendix 3). (iv) Pereoglou, F., Banks, S.C., MacGregor, C., Ford, F., Wood, J., Lindenmayer, D.B. How does fire history influence demographic parameters and genetic patterns in an early successional specialist small mammal? Ecology and Evolution, Submitted.
The research, analysis and writing in the above publications was substantially (>90%) my own work. I am also the corresponding author for each publication.
Felicia Pereoglou Date: 14 March 2016
Co-author acknowledgement:
David B. Lindenmayer Date: 14 March 2016
Pereoglou, F., Population processes in an early successional heathland 21 Chapter 1: Introduction
Sam Banks Date: 14 March 2016
Jeff T. Wood Date: 14 March 2016
Christopher MacGregor Date: 14 March 2016
Fred Ford Date: 14 March 2016
Pereoglou, F., Population processes in an early successional heathland 22 Chapter 1: Introduction
1.5 References
Bradstock, R. A., Williams, J. & Gill, A. M. (2012) Flammable Australia: The Fire Regimes and Biodiversity of a Continent. Cambridge University Press, Cambridge. Breed, B. & Ford, F. (2007) Native Mice and Rats, CSIRO Publishing, Collingwood. Catling, P. C. (1986) Rattus lutreolus, colonizer of heathland after fire in the absence of Pseudomys species. Australian Wildlife Research, 13, 127-139. Catling, P. C. & Newsome, A. E. (1981) Responses of the Australian vertebrate fauna to fire: an evolutionary approach. Fire and the Australian Biota (eds A. M. Gill, R. H. Groves & I. R. Noble), pp. 273 - 310. Australian Academy of Science, Canberra. Connell, J. H. & Slatyer, R. O. (1977) Mechanisms of succession in natural communities and their role in community stability and organization. American Naturalist, 111, 1119- 1144. Delmoral, R. & Bliss, L. C. (1993) Mechanisms of primary succession: insights resulting from the eruption of Mount St Helens. Advances in Ecological Research, 24, 1-66. Dickman, C. R., Lunney, D. & Matthews, A. (2000) Ecological attributes and conservation of native rodents in New South Wales. Wildlife Research, 27, 347-355. Flannigan, M. D., Krawchuk, M. A., de Groot, W. J., Wotton, B. M. & Gowman, L. M. (2009) Implications of changing climate for global wildland fire. International Journal of Wildland Fire, 18, 483-507. Fox, B. J. (1982) Fire and mammalian secondary succession in an Australian coastal heath. Ecology, 63, 1332-1341. Fox, B. J. (2008) Eastern chestnut mouse, Pseudomys gracilicaudatus. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 634-635. Reed New Holland, Sydney, Australia. Fox, B. J. & Fox, M. D. (1981) A comparison of vegetation classifications as descriptors of small mammal habitat preference. Vegetation Classification in Australia: Proceedings of a Workshop Sponsored by CSIRO Division of Land Use Research, Canberra, October 1978 (eds A. N. Gillison & D. J. Anderson), pp. 166-180. CSIRO, Canberra. Fox, B. J. & Kemper, C. M. (1982) Growth and development of Pseudomys gracilicaudatus (Rodentia: Muridae) in the laboratory. Australian Journal of Zoology, 30, 175-185. Fox, B. J., Taylor, J. E. & Thompson, P. T. (2003) Experimental manipulation of habitat structure: a retrogression of the small mammal succession. Journal of Animal Ecology, 72, 927-940. Gill, A. M. (1975) Fire and the Australian flora: a review. Australian Forestry, 38, 4-25. Gill, A. M. & Catling, P. C. (2002) Fire regimes and biodiversity of forested landscapes of southern Australia. Flammable Australia: The Fire Regimes and Biodiversity of a
Pereoglou, F., Population processes in an early successional heathland 23 Chapter 1: Introduction
Continent (eds R. A. Bradstock, J. E. Williams & A. M. Gill), pp. 94-124. Cambridge University Press, New York. Haering, R. & Fox, B. J. (1995) Habitat utilization patterns of sympatric populations of Pseudomys gracilicaudatus and Rattus lutreolus in coastal heathland: a multivariate analysis. Australian Journal of Ecology, 20, 427-441. Higgs, P. & Fox, B. J. (1993) Interspecific competition: a mechanism for rodent succession after fire in wet heathland. Australian Journal of Ecology, 18, 193-201. Keith, D. A., McCaw, W. L. & Whelan, R. J. (2002) Fire regimes in Australian heathlands and their effects on plants and animals. Flammable Australia: the fire regimes and biodiversity of a continent (eds R. A. Bradstock, J. E. Williams & A. M. Gill), pp. 199- 237. Cambridge University Press, New York. Lee, A. K., Baverstock, P. R. & Watts, C. H. S. (1981) Rodents - the late invaders. Ecological Biogeography of Australia (ed A. Keast), pp. 1521-1553. Dr W Junk Publishers, The Hague, London. Luo, J. & Fox, B. J. (1994) Diet of the Eastern chestnut mouse (Pseudomys gracilicaudatus) II Seasonal and successional patterns. Wildlife Research, 21, 419-431. Luo, J., Fox, B. J. & Jefferys, E. (1994) Diet of the Eastern Chestnut Mouse (Pseudomys gracilicaudatus) I Composition, diversity and individual variation. Wildlife Research, 21, 401-417. Mahoney, J. A. & Posamentier, H. (1975) The occurrence of the native rodent, Pseudomys gracilicaudatus (Gould, 1845) [Rodentia: Muridae], in New South Wales. Australian Mammalogy, 1, 333-346. Meek, P. D. & Triggs, B. (1997) Extension to the range of the eastern chestnut mouse, Pseudomys gracilicaudatus, to Jervis Bay, New South Wales. Australian Mammalogy, 20, 123-125. Miller, A. D., Roxburgh, S. H. & Shea, K. (2011) How frequency and intensity shape diversity- disturbance relationships. Proceedings of the National Academy of Sciences, 108, 5643- 5648. Monamy, V. & Fox, B. J. (2000) Small mammal succession is determined by vegetation density rather than time elapsed since disturbance. Austral Ecology, 25, 580-587. Monamy, V. & Fox, B. J. (2010) Responses of two species of heathland rodents to habitat manipulation: vegetation density thresholds and the habitat accommodation model. Austral Ecology, 35, 334-347. Morris, D. W., Fox, B. J., Luo, J. & Monamy, V. (2000) Habitat-dependent competition and the coexistence of Australian heathland rodents. Oikos, 91, 294-306. Rossi, F., Forster, R. M., Montserrat, F., Ponti, M., Terlizzi, A., Ysebaert, T. & Middelburg, J. J. (2007) Human trampling as short-term disturbance on intertidal mudflats: effects on macrofauna biodiversity and population
Pereoglou, F., Population processes in an early successional heathland 24 Chapter 1: Introduction dynamics of bivalves. Marine Biology, 151, 2077-2090. Sousa, W. P. (1979) Disturbance in intertidal boulder fields: the non-equilibrium maintenance of species diversity. Ecology, 60, 1225-1239. Steffen, W., Burbidge, A. A., Hughes, L., Kitching, R., Lindenmayer, D. B., Musgrave, W., Stafford Smith, M. & Werner, P. (2009) Australia's Biodiversity and Climate Change, CSIRO Publishing, Melbourne. Strahan, R. (2008) The Mammals of Australia. Reed, Sydney. Swanson, M. E., Franklin, J. F., Beschta, R. L., Crisafulli, C. M., DellaSala, D. A., Hutto, R. L., Lindenmayer, D. B. & Swanson, F. J. (2011) The forgotten stage of forest succession: early-successional ecosystems on forest sites. Frontiers in Ecology and the Environment, 9, 117-125. Velle, L. G. & Vandvik, V. (2014) Succession after prescribed burning in coastal Calluna heathlands along a 340-km latitudinal gradient. Journal of Vegetative Science, 25, 546- 558. Watts, C. H. S. & Aslin, H. J. (1981) The Rodents of Australia, Angus & Robertson Publishers, Sydney. Whelan, R. J. (1995) The Ecology of Fire, Cambridge University Press, Cambridge. Williams, A. A. J., Karoly, D. J. & Tapper, N. (2001) The sensitivity of Australian fire danger to climate change. Climatic Change, 49, 171-191.
Pereoglou, F., Population processes in an early successional heathland 25
Chapter 2: Refuge site selection
Refuge site selection
Published as:
Pereoglou, F., MacGregor, C., Banks, S.C., Ford, F., Lindenmayer, D.B. (2011) Refuge site selection by the eastern chestnut mouse in recently burned heath Wildlife Research 38(4): 290-298. doi:10.1071/WR11007
Manuscript: Appendix 1
Pereoglou, F., Population processes in an early successional heathland 26 Chapter 2:Refuge site selection
2.1 Abstract Temporal reduction in shelter is an indirect primary impact of fire. Removal of animal refugia has implications for shelter site selection and fidelity – factors that have been largely overlooked in studies of Australian rodent fauna. This information is critical for guiding species conservation and appropriate land management including prescribed burning practices. We sought to determine which features of burned heath were selected as shelter sites by the eastern chestnut mouse, whether there was sex and/or seasonal variation in shelter site selection and whether we could identify primary refugia. We completed a radiotelemetry study to identify diurnal refuge sites and compare habitat attributes with those of a matched set of control sites. We then used habitat features and fidelity parameters to classify refuge site use. We found the eastern chestnut mouse selected shelters with the presence of specific structures and had taller, denser vegetation than randomly selected control sites. There were no differences in habitat selection between the sexes. Shelter sites in the non-breeding season had greater vegetation density compared to those used in the breeding season. In the breeding season, the eastern chestnut mouse showed no evidence of increased fidelity to particular refuges. Vegetation density in winter was the best predictor of a primary refuge compared to whether or not an animal returned to a shelter site or the amount of time spent in a shelter site. Mice were ephemeral and non-gregarious in their refuge use. There was some evidence for inheritance of refuge sites from a parent, as well as inter-season shelter site fidelity. The eastern chestnut mouse selected refugia that had habitat attributes offering maximum protection. Seasonality in refuge site selection is likely to reflect the reproductive and environmental trade-offs in critical resources during different seasons. The maintenance of multiple, rarely occupied shelters by the eastern chestnut mouse is consistent with data for other mammals. Fire management should ensure retention of vegetation structure on the ground layer, dense habitat patches in burnt areas, and be carefully planned during the winter season to maintain shelter and refuge sites to assist population persistence.
2.2 Introduction The response of small mammal populations to fire has been studied extensively in Australia (Newsome et al. 1975; Cockburn et al. 1981; Fox 1982; Newsome and Catling 1983; Catling 1986; Thompson et al. 1989; Catling 1991; Masters 1993; Letnic 2003; Letnic et al. 2004; Lindenmayer et al. 2008). A pattern of post-fire irruption followed by decline has been observed for some Australian native rodents (Whelan et al. 2002). The eastern chestnut mouse (Pseudomys gracilicaudatus) is one example. Existing evidence suggests the species is a pyric specialist with peak population density in recently burned heathland and a subsequent decline in abundance with time since fire (Fox 1982; Higgs and Fox 1993; Thompson and Fox 1993; Monamy and Fox 2000; Fox et al. 2003; Monamy and Fox 2010).
Pereoglou, F., Population processes in an early successional heathland 27 Chapter 2:Refuge site selection
The eastern chestnut mouse occurs at low densities when present, is rare or extinct in many parts of its range (Breed and Ford 2007) and is classified as vulnerable under the Threatened Species Conservation Act 1992 (NSW). Inappropriate fire regimes including fire suppression in areas supporting populations of the eastern chestnut mouse may affect the species’ persistence and viability. However, the mechanisms underlying the species’ fire response are unknown. Terrestrial small mammals can be affected by fire through direct mortality, and indirectly through changes in resource availability (Whelan et al. 2002). Studies of the eastern chestnut mouse’s dietary requirements indicate the species is a generalist herbivore with an opportunistic foraging strategy able to utilise a diverse food resource (Luo and Fox 1994; Luo et al. 1994). A diverse diet is a basic requirement for rapid colonisation of recently burnt areas and is an essential attribute for a fire opportunist to take advantage of all available food resources (Luo et al. 1994). Given this, it is unlikely that depletion of the food resource alone would be a determinant of the species’ fire response. The role of animal refugia in facilitating population recovery and persistence after a fire is poorly understood for Australian rodent fauna. A reduction in available shelter sites is one of the primary impacts of fire on populations (Newsome and Catling 1983; Friend 1993) and may have multiple consequences including distributional, demographic and behavioural change (Sutherland and Dickman 1999). The presence, size and location of refugia are landscape attributes that may influence the processes of mortality, colonisation, post-fire survival, establishment, reproduction and population increase (Whelan et al. 2002). Daytime refugia and shelter sites are critical for surviving disturbance events (Braithwaite 1990), environmental extremes, predation and for providing security while weaning young (Frank and Layne 1992). Identification of habitat components that might represent refuges and the question of how important they are for maintenance of post-fire populations requires additional study (Keith et al. 2002). The rate of reappearance after fire is dependent on a species’ mobility, dispersal capacity, dietary specialisation and habitat preferences (Keith et al. 2002), including the level of protection provided by refugia during and after a fire. An assumption of survival within burnt areas would lead us to predict rapid reappearance and population growth (Keith et al. 2002). Recruitment from residual animals after disturbance has been shown to be important (Lindenmayer et al. 2005; Banks et al. 2011). This suggests refugia may play a critical role in a species’ ability to survive fire and the subsequent population response. The identification, distribution, and protection of animal shelter sites and refuges should be an essential component of conservation strategies in fire-prone landscapes. In this paper, we document the results of a radiotelemetry study of daytime refuge habitat requirements and nest fidelity of the eastern chestnut mouse. We sought to answer the following questions:
Pereoglou, F., Population processes in an early successional heathland 28 Chapter 2:Refuge site selection
(i) Does the eastern chestnut mouse preferentially select shelter sites based on specific habitat attributes? We expected that mouse-selected shelter sites would be characterised by denser, taller vegetation than control sites. (ii) Are there sex differences or seasonal variation in shelter site selection and fidelity? We hypothesised the female mice would select shelter sites that offer optimal cover and protection during summer and would show greater shelter site fidelity than males. We expected these differences to be driven by different requirements of the sexes during the breeding season – care of offspring in females versus proximity to mates for males. We predict there would be no difference between male and female mice in shelter site selection and fidelity during winter. (iii) Is there a primary–secondary model of refuge use? We hypothesised that animals would show increased fidelity to shelters characterised by particular attributes including the presence of a mouse-made nesting structure (Wolfe 1970; Klein and Layne 1978; Hartung and Dewsbury 1979). These primary refuges would be used more frequently and would have habitat attributes that distinguish them from secondary shelters. (iv) Does the eastern chestnut mouse share shelter sites? We expected that individuals would not share shelter sites, instead likely to be solitary in their nesting habits. The closely related western chestnut mouse (P. nanus) also has a dispersed, solitary social organisation (Breed and Ford 2007) and other Pseudomys species have been shown to engage in agonistic nest defence (Happold 1976). Identification of small mammal shelter requirements and an understanding of refuge site fidelity and nesting behaviour could be used by land managers when designing appropriate burning regimes or promoting post-fire population recovery. For threatened species, knowledge of nest requirements and daytime refuge habitat and use is fundamental for conservation efforts.
2.3 Materials and methods 2.3.1 Study sites We selected six sites for study in Booderee and Jervis Bay national parks in south-eastern Australia (Fig. 2.1). These adjoining reserves are located approximately 200 km south of Sydney on the south coast of New South Wales. We based site selection on survey data of Townley (2007), Lindenmayer et al. (2008) and a pilot study indicating the presence of the eastern chestnut mouse at sufficiently high densities to enable attachment of radio-transmitters to at least two individuals per site. We classified study sites as heath (defined by Taws (1997) as vegetation dominated by shrubs with small narrow leaves), generally less than 2 m tall and varyingly dominated by Banksia ericifolia and Sprengelia incarnata, with Allocasuarina distyla, Isopogon anemonifolius and Hakea teretifolia also being common. Each of our six field sites had been burnt by wildfire within the last six years.
Pereoglou, F., Population processes in an early successional heathland 29 Chapter 2:Refuge site selection
Figure 2.1: Location of study sites in Booderee and Jervis Bay national parks
2.3.2 Trapping and radio-collaring We completed trapping and radio-collaring during the 2008–2009 summer breeding season (November-March) and the winter 2009 non-breeding season (June-August). We captured animals using grids of type A Elliot traps open for four consecutive nights using a bait of rolled oats, peanut butter and fennel seed. During winter, we insulated traps with non-absorbent cotton wool and covered them with a plastic sleeve. Our grid arrangement comprised lines at 20 m intervals with traps 10 m apart. We provide a summary of field site details, trapping and radio- collaring data (Table 2.1). We assessed the weight, sex, age cohort and breeding condition of captured animals. We marked mice for permanent identification using Trovan ID100 implantable transponders (Microchips Australia Pty Ltd, Keysborough, Victoria, Australia). We selected mice for radio- collar attachment if they were (i) an adult, (ii) weighed >70 g to ensure collar weight was less than 5% of an individual’s body weight, and (iii) not visibly pregnant. We fitted individuals with a single-stage transmitter on a replaceable cable tie weighing 2.5-3.5 g (Sirtrack Limited, Havelock North, Hawkes Bay, New Zealand) in the field before release at the point of capture. We tracked animals to their diurnal shelter site daily for a period of one month using a Yagi hand-held antenna and a Communications Specialist R1000 receiver (Sirtrack Limited, Havelock North, Hawkes Bay, New Zealand). Transmission was highly directional at close
Pereoglou, F., Population processes in an early successional heathland 30 Chapter 2:Refuge site selection
range allowing us to pinpoint animals to within 1 m2. We numbered and permanently marked shelter sites, and recorded location (GPS coordinates).
Table 2.1: Trapping and radio-collaring site summary Location Site Area Trap No. of No. of ♂ ♀ Trap No. of No. of ♂ ♀ (ha) nights captures collars nights captures collars summer fitted winter fitted BNP 7-44B 2.40 960 18 4 1 3 480 10 6 (1) 2 4 BNP 7-I 2.05† 820 17 6 4 2 400 8 6 (2) 2 4 BNP 7-2 2.00 400 19 4 3 1 400 13 6 (1) 4 2 BNP 7-46 3.25† 1280 36 7 2 5 480 7 5 (0) 4 1 BNP 7-H 2.20† 880 20 5 2 3 400 5 3 (1) 2 1 JNP 7-ST1 2.50 500 7 3 1 2 500 11 3 (1) 1 2 Total 14.40 4840 117 29 13 16 2660 54 29 (6) 15 14
2.3.3 Assessment of shelter sites We assessed shelter sites for vegetation density, height and structure. We measured vegetation density with a light meter (in lux) at 20 cm intervals of a 1 m profile (in duplicate) through the centre of the shelter site. We used an umbrella to shade the light meter from direct sunlight to enable assessment to occur in diffuse light conditions on any given day or time of day. We also recorded the maximum height of vegetation and the structure(s) providing the greatest cover. We searched shelter sites for the presence of a mouse-made nest structure. For each shelter site, we also assessed a control site using the identical procedure. We located control sites at a random compass bearing and a random distance between 2 and 20 m from a known shelter site.
2.3.4 Data analysis We used paired t-tests to determine whether shelter sites differed from controls in vegetation density (estimated as mean lux at 0, 20, 40, 60, 80 and 100 cm), vegetation height and total structure(s) present. We used two-sample binomial tests (Collett 2002) to determine whether mice preferentially selected the presence of structures of different types. To identify effects of sex and/or season in the selection of shelter sites, we used linear mixed models (McCulloch et al. 2008) with the response variables being vegetation height and log- transformed standardised light intensity (as a surrogate for vegetation density, calculated as the proportion of light reaching the ground layer relative to what was available at 1 m). We fitted animal identifier and site as random terms to evaluate individual animal and site variation. We excluded control sites from this analysis. Using control sites only, we investigated vegetation structure for its effect on standardised light intensity (where standardised light intensity approaching zero would approximate greater ‘cover’). We fitted all possible regression models and ranked them using the Schwarz information criterion (SIC) (Schwarz 1978). The candidate variables were the representative structural categories: Xanthorrhea, buttongrass, dead stems, shrub and no cover.
Pereoglou, F., Population processes in an early successional heathland 31 Chapter 2:Refuge site selection
We considered a primary refuge to be one with a nest structure present. Mouse-made nest structures consisted of finely processed dried grass or sedge woven into either (i) a pad approximately 10 cm in diameter, or (ii) a sphere approximately 10 cm in diameter and height with an inconspicuous opening approximately 2 cm wide. We fitted generalised linear models (GLMs) (McCullagh and Nelder 1989) assuming a Bernoulli response with a logit link function for the probability of occurrence of a nest. The candidate predictors were sex, season, habitat attributes (vegetation density=log-transformed standardised light intensity, vegetation height and structure) and fidelity parameters (re-use, count of fixes standardised by number of radiotracking days). We ranked all possible models using SIC.
2.4 Results 2.4.1 Trapping and radio-tracking We attached radio-collars to 58 of the 140 individuals of eastern chestnut mouse that we captured. We summarised data on the number of mice captured and radio-collared at each site in each season (Table 2.1). On average, we tracked each animal 22 ± 1.53 times (mean ± s.e.) to diurnal shelter sites. The overall success rate of radio-tracking individuals for >20 consecutive fixes was 67 % with the remaining animals tracked less intensively due to predation, radio- collar loss or signal failure.
2.4.2 Shelter site selection We identified 453 unique shelter sites from 1300 tracking fixes. On average, each animal had 8 ± 0.5 shelter sites, spent 2.5 ± 0.1 days (total consecutive and non-consecutive fixes) in a shelter site and re-visited shelter sites 0.8 ± 0.1 times (defined as the number of times a mouse returned to a previously identified shelter site). Above ground nest structures were present in 15% of shelter sites. We did not locate any below ground burrow systems. We found shelter sites had lower lux values (=more dense vegetation) up to 60 cm above the ground (Fig. 2.2) and taller vegetation with more structure compared to control sites (Table 2.2). The type of structure present also differed between shelter and control sites with Xanthorrhea, buttongrass and dead stems preferentially selected by the eastern chestnut mouse (Table 2.3). These structures were significant predictors of decreasing standardised light intensity when we analysed control site data separately (Table 2.4). Irrespective of individual animal and site variation, male and female mice did not differ in the standardised light intensity of sites they selected for sheltering but showed a tendency to choose sites with lower standardised light intensity during winter (F1,88=9.6, p=0.003, Fig. 2.3). The height of shelter sites did not differ significantly between sexes or across seasons.
Pereoglou, F., Population processes in an early successional heathland 32 Chapter 2:Refuge site selection
Figure 2.2: Light intensity measured in lux at 20 cm intervals of a 1m profile through eastern chestnut mouse shelter sites (solid line) and paired control sites (broken line). Loss of light is a surrogate for vegetation density. Bars indicate mean s.e. Values were significantly different at the 0 – 0.6 m height categories.
Table 2.2: Means ( s.e.), t-statistics and significance levels for differences in height and total number of structures present between paired shelter and control sites Variate Shelter Control T P Height (m) 0.869 ± 0.012 0.785 ± 0.013 4.99 <0.001 Total structures 1.589 ± 0.028 1.243 ± 0.023 10.43 <0.001
Table 2.3: Proportion of successes, test statistic and significance level for differences in the types of structures present in shelter and control sites Structure type Shelter Control s.e. difference P Xanthorrhea 0.496 0.211 0.029 <0.001 Buttongrass 0.068 0.008 0.123 <0.001 Dead stems 0.380 0.257 0.030 <0.001 Shrub 0.384 0.696 0.031 <0.001 No cover 0.255 0.458 0.030 <0.001
Table 2.4: Multiple regression statistics for a test of the effect of type of structure on standardised light intensity. Analysis excluded shelter sites due to mouse-selection bias Parameter Estimate s.e. t-statistic P Constant -1.25 0.025 -49.98 <0.001 Xanthorrhea -0.39 0.047 -8.28 <0.001 Buttongrass -1.08 0.205 -5.24 <0.001 Dead stems -0.13 0.044 -2.91 0.004
Pereoglou, F., Population processes in an early successional heathland 33 Chapter 2:Refuge site selection
Figure 2.3: Log-scale standardised light intensity (sLI) of male and female eastern chestnut mouse shelter sites ( s.e.) selected during different seasons.
2.4.3 Refuge use and fidelity We were able to predict a primary refuge (i.e. those scored present for a nest structure) based on log-transformed standardised light intensity and season (Table 2.5). We found primary refugia were characterised by vegetation that was denser than secondary shelters (i.e. mouse- selected shelter sites that did not have a nest structure present). This relationship was significant in the non-breeding season indicating mice preferentially chose to construct a nest in a shelter site if it comprised very dense vegetation (Table 2.5). During the breeding season, mice did not appear to be as selective and any given shelter site was equally likely to have a nest structure. The eastern chestnut mouse was generally a solitary nester. We observed 1165 instances of solitary refuge use. We observed only very occasionally shelter site sharing and confirmed one record of cohabitation. The most common types of sharing we recorded were (i) where two different mice used the same shelter site during different seasons (mice ‘inherited’ the shelter site from a previous occupant – 14 observations); or (ii) where two different mice used the same shelter site at different times during the same season of radio-tracking (sequential shelter site use – 14 observations; six in summer and nine in winter). We observed a male-female pair of radio-collared mice in the same shelter site on one occasion for 14 contiguous daily fixes during winter. This was the only record of cohabitation. We were unable to determine the genetic relationship (if any) between these two mice. We also recorded seven observations of long term shelter site fidelity where the same individual was radio-tracked in multiple seasons and used
Pereoglou, F., Population processes in an early successional heathland 34 Chapter 2:Refuge site selection the same shelter sites in both periods. This is potentially more common than recorded here due to the small number of mice we re-collared and radio-tracked during both seasons.
Table 2.5: Parameters for predicting presence of a primary refuge. Season refers to the difference between summer breeding and winter non-breeding seasons Parameter Estimate s.e. Statistic P Constant -3.69 0.90 -4.09 <0.001 Log10 sLI -0.63 0.39 -1.61 0.108 STD count of fixes 0.74 1.25 0.59 0.553 Season -3.22 1.42 -2.27 0.023 Log10 sLI * season -1.55 0.59 -2.60 0.009 STD count of fixes * season 3.58 1.86 1.93 0.054
2.5 Discussion 2.5.1 Refuge site selection Selection of appropriate refuge sites represents a substantial investment of time and energy by an animal. Animals spend periods of inactivity in these locations and shelter and raise their young in them. Therefore, the construction of nests and selection of refuge sites should afford the best protection from predation for both adults and offspring to maximise fitness (Stokes 1995). The eastern chestnut mouse preferentially selected refuge sites with particular habitat attributes. We found shelter sites were selected that had taller, denser vegetation compared to the surrounding habitat. Sites with multiple structures, such as Xanthorrhea, Gymnoschoenus (buttongrass) and the collapsed dead shrubs were also often selected. We suggest these features have a combined effect in providing safety from trampling and predation, insulation against unfavourable climatic conditions, access to foraging areas and protection from aggressive interactions with competitors. For the eastern chestnut mouse, the selection of dense, protected areas for shelter may reflect competitor avoidance. Previous studies indicate competitive exclusion by the swamp rat (Rattus lutreolus) on eastern chestnut mouse access to food resources (Luo and Fox 1995; Luo et al. 1994) and preferred habitat (Higgs and Fox 1993; Morris et al. 2000). In our study, the swamp rat was absent from four of the six study sites. Refugia selected by the eastern chestnut mouse in the two sites where swamp rats were present, were not consistently different from those selected in the sites where the swamp rat was absent. From the results of our study, it is impossible to determine the extent of competition for shelter. However, dense cover has been shown to be important for the swamp rat (Catling 1986; Haering and Fox 1995; Monamy and Fox 2000; Fox and Monamy 2007). The height of vegetation and areas of dense cover have been shown to be an important factor in the choice of refuge location for other terrestrial rodent species (Frank and Layne 1992; Stokes 1995; Gray et al. 1998; Wagner et al. 2000). The silky mouse’s (P. apodemoides) preference for low, dense
Pereoglou, F., Population processes in an early successional heathland 35 Chapter 2:Refuge site selection vegetation and the availability of desert banksia (B. ornata) was considered a major influence on the distribution and location of refugia purposes (Cockburn 1981a).
2.5.2 Refuge site use and fidelity Daytime refuges can include both long-term shelter sites and temporary resting places (Wolff and Hurlbutt 1982). Given the multiple functions of a refuge site, we expected to see a primary–secondary model of refuge use. Refuges offering the best protection are likely to be used most frequently and for long periods of time. The acquisition of refugia has a spatial element. To take advantage of dispersed food resources, animals might have many, less frequently used refuge sites (Ward 1978). We found evidence for hierarchical refuge selection. We considered the presence of a nesting structure to be indicative of a primary refuge. The number of days spent in a shelter and whether a shelter site was re-visited during the course of the study were not significant predictors of a primary refuge. However, we did find a seasonal effect with the density of the vegetation in winter being the best predictor of whether a shelter site would be selected for building a nesting structure. Even though females produced litters during the breeding season, they were no more likely to build nesting structures than males. Any shelter site in the breeding season had an equal likelihood of being a primary refuge. We suggest this may be a result of (i) reproductive constraints including competition for mates, (ii) predator avoidance during the summer months, (iii) competition for foraging resources among females weaning litters, and an (iv) an insufficient survey of breeding females because we did not attach radio-collars to visibly pregnant animals. Retaining primary refuge sites over winter is essential for animals to regulate against thermal extremes and ensure access to foraging resources while competing with other adult animals, sub-adults and juvenile offspring from the previous breeding season. Seasonality in shelter site use has been recorded for other rodent species (Madison et al. 1984; Wolton and Trowbridge 1985; Bubela and Happold 1993) and was attributed to environmental factors (i.e. shift in shelter site location dependent on ambient temperature (Madison et al. 1984)) and functionality (i.e. different refuges for weaning litters (Morzillo et al. 2003)) compared to refugia solely for sheltering.
Many mammals defend multiple, rarely occupied refuge sites (Happold 1976; Cockburn 1981a; Wolff and Hurlbutt 1982; Lindenmayer et al. 1996; Hall and Morrison 1997; Moro and Morris 2000; Woods and Ford 2000; Morzillo et al. 2003; Sanecki et al. 2006; Hinkelman and Loeb 2007; Ebensperger et al. 2008). Refuge fidelity and site attachment has links to social organisation and animal behaviour. The eastern chestnut mouse was transient in their refuge use. Our observations indicate the eastern chestnut mouse was most commonly a solitary nester. Solitary sheltering might reflect sparse food resource availability or a territorial social organisation (West and Dublin 1984; Ostfeld 1985). There was some evidence for sharing of refuge sites and long-term fidelity to particular sites. Long-term site fidelity is potentially more common than was possible to record in this study due to the small number of animals we were
Pereoglou, F., Population processes in an early successional heathland 36 Chapter 2:Refuge site selection re-collared during both seasons. The most commonly recorded instances of shelter site sharing were temporal segregation or inheritance of a shelter site from a previous occupant. There was only one record of a pair-bond and this was observed during winter. The benefits of refuge site fidelity and defence include: (i) highlighting ownership of, and defence of a shelter site against conspecifics; (ii) reduced parasite burden; (iii) avoiding predators learning patterns of emergence behaviour; and (iv) periodic change in foraging area utilised (Lindenmayer et al. 1996). Familiarity with an area and a known location of proven shelter is probably crucial to survival when exploring unknown territory (Gray et al. 1998). Shelter site replacement (occupation by alternate animals) is not necessarily a random event (Haim and Rozenfeld 1995). We propose that shelter sites used by different individuals of the eastern chestnut mouse in different seasons may result from parent-offspring or sibling relationship between animals. Bonds resulting from genealogical relationships (parent-offspring and sibling bonds) are not uncommon for pseudomyine rodents (Happold 1976). The silky mouse has been recorded from communal burrows in groups comprising a breeding pair with one or more litters of varying developmental stages (Cockburn 1981b). The Hastings River mouse (P. oralis) has high shelter site fidelity (Townley 2000) and there is some evidence to suggest offspring inherit the maternal range (S. Townley pers. comm..). Further research is needed to confirm if this is the case in the eastern chestnut mouse and it would be of interest to determine whether the relationship between animals is a significant determinant of spatial patterns in sheltering.
2.5.3 Prescribed burning practices Fire management in areas supporting populations of the eastern chestnut mouse should ensure the retention of tall, dense vegetation and structural features in burnt landscapes. We speculate that as time since fire increases, eastern chestnut mouse shelter sites have the potential to become a limiting resource. Vegetation regeneration in heathland would result in a sparse ground layer and in the complete absence of fire we could expect fewer structural components to be available. That is: (i) the current collapsed shrub layer would decompose; (ii) although a long-lived, slow-growing species, the thatched, dead leaves of unburned Xanthorrhea would no longer provide cover at ground level as plants grew taller (Gill and Ingwersen 1976); and (iii) woody species would dominate or eliminate the herbaceous plants increasing in biomass and height (Ingwersen 1977; Specht 1981; Keith et al. 2002). Conversely, immediately after a fire, it is also reasonable to expect shelter sites and refugia to be limiting, particularly following a high intensity burn. In contrast to other Pseudomys species (ash-grey mouse P. albocinereus (Happold 1976), silky mouse (Cockburn 1981a), smoky mouse P. fumeus (Woods and Ford 2000), New Holland mouse P. novaehollandiae (Kemper 1981), and heath rat P. shortridgei (Taylor and Horner 1972; Happold 1976)) and Fox (1995), we did not find the eastern chestnut mouse to use a below ground burrow complex. Burrows provide a more fire resilient shelter (Happold 1976; Kemper 1981; Friend 1993; Sutherland and Dickman 1999; Whelan et al.
Pereoglou, F., Population processes in an early successional heathland 37 Chapter 2:Refuge site selection
2002). With the complete removal of above ground vegetation, the ability of refugial populations to survive in unburnt patches would be a critical factor affecting population persistence (Friend 1993; Whelan 1995). In a manipulative experiment designed to promote small mammal responses to a simulated fire, there was no effect of a reduction in cover on the eastern chestnut mouse (Fox et al. 2003). However, the spatial scale and structural effect of wildfire is likely to be different from that of experimental clipping, which, for instance contained intact patches of vegetation within treatment plots (i.e. buttongrass). Further studies are necessary to determine whether the changes in vegetation regrowth after fire affect habitat preferences of shelter sites and refugia.
Assuming the occurrence of fire in heathland is important for retention of habitat features selected by the eastern chestnut mouse for refugia, appropriate timing of prescribed burning practices may be important. Within mouse-selected refuge sites, those with denser vegetative cover were used for over-wintering. These shelters were significantly more likely to have a nest structure present. The eastern chestnut mouse builds an above ground surface nest comprised of dried sedge or grassy material woven into a pad or spherical structure. These are highly flammable refuges. Protection from fire by such a shelter would be limited and the likelihood of direct mortality by fire would be very high. The eastern chestnut mouse nest structure resembles that of the eastern bristle bird (Dasyornis brachypterus); a co-occurring heathland species which has also been found absent immediately after wildfire from sites that were previously occupied (Lindenmayer et al. 2009). The only other Pseudomys species known exclusively to build an above ground grass nest is the western chestnut mouse (P. nanus) which is also now extinct from the southern part of the species’ historical range and is rare or extinct throughout current distribution (Strahan 1995; Breed and Ford 2007). Management practices that allow burning during winter months would remove valuable primary refuge sites and may result in the post- fire residual eastern chestnut mouse population expending greater energy to re-establish over- wintering nesting structures at a cost of reduced foraging capacity.
2.5.4 Conclusion The eastern chestnut mouse preferentially selects relatively taller, denser vegetation (up to 1 m) with structural features present at the ground layer for refuge. Refugia are highly flammable and do not provide protection from fire. These shelter site requirements have implications for land management and prescribed burning practices. There is some evidence for a primary-secondary model of refuge use. Primary refuge sites had a nesting structure present and were the more dense shelter sites selected by the eastern chestnut mouse. Any shelter had an equal likelihood of being a primary or secondary refuge during the summer breeding season but dense vegetation was the best predictor of a primary refuge for over-wintering animals. There are likely to be several reasons for this seasonal aspect of shelter site use including reproductive constraints and intra-specific competition for reproductive resources in the summer breeding
Pereoglou, F., Population processes in an early successional heathland 38 Chapter 2:Refuge site selection season and environmental constraints including regulation against thermal extremes in the absence of huddling behaviour in winter. The maintenance of multiple, rarely occupied shelters is consistent with data for other mammals and is likely to be the result of territorial defence, access to dispersed resources and predator avoidance. The eastern chestnut mouse is mostly solitary in its sheltering habits but there is some evidence for short-term sharing, temporal segregation in the use of the same sites, or familial inheritance in the longer time frame.
2.6 Acknowledgements We most gratefully acknowledge the Wreck Bay Aboriginal Community and the Wandandian People of the Dharawal and Dhurga language groups on whose lands this study was conducted. For valuable field assistance, advice and discussions, we thank Wendy Hartman, Alfred Wellington, Tony Carter, Geoff Kay, Steve Phillips, Marama Hopkins, Nick Dexter, Karen Viggers, Sally Townley, Les Mitchell and Martin Westgate. This research was funded by the Norman Wettenhall Foundation and the MA Ingram Trust and conducted under The Australian National University Animal Experimentation Ethics Committee protocol C.RE.49.08, NSW National Parks and Wildlife Service Scientific Licence S12690 and Booderee National Park Permit BDR08/00009.
2.7 References Banks SC, Dujardin M, McBurney L, Blair D, Barker M, Lindenmayer DB (2011) Starting points for small mammal population recovery after wildfire: recolonisation or residual populations? Oikos 120(1), 26-37. Braithwaite RW (1990) Australia's unique biota: implications for ecological processes. Journal of Biogeography 17(4-5), 347-354. Breed B, Ford F (2007) 'Native Mice and Rats.' (CSIRO Publishing: Collingwood) Bubela TM, Happold DCD (1993) The social organization and mating system of an Australian sub-alpine rodent, the broad-toothed rat Mastacomys fuscus Thomas. Wildlife Research 20(4), 405-417. Catling PC (1986) Rattus lutreolus, colonizer of heathland after fire in the absence of Pseudomys species. Australian Wildlife Research 13(2), 127-139. Catling PC (1991) Ecological effects of prescribed burning practices on the mammals of southeastern Australia. In 'Conservation of Australia's Forest Fauna.' Ed. D Lunney). (Royal Zoological Society of NSW: NSW) Cockburn A (1981a) Diet and habitat preference of the silky desert mouse Pseudomys apodemoides (Rodentia). Australian Wildlife Research 8(3), 475-97.
Cockburn A (1981b) Population processes of the silky desert mouse Pseudomys apodemoides (Rodentia) in mature heathlands. Australian Wildlife Research 8(3), 499-514.
Pereoglou, F., Population processes in an early successional heathland 39 Chapter 2:Refuge site selection
Cockburn A, Braithwaite RW, Lee AK (1981) The response of the heath rat, Pseudomys shortridgei, to pyric succession - a temporally dynamic life-history strategy. Journal of Animal Ecology 50(3), 649-666. Collett D (2002) 'Modelling Binary Data.' 2nd edn. (Chapman and Hall: London) Ebensperger LA, Sobrero R, Campos V, Giannoni SM (2008) Activity, range areas, and nesting patterns in the viscacha rat Octomys mimax. Journal of Arid Environments 72(7), 1174- 1183. Fox BJ (1982) Fire and mammalian secondary succession in an Australian coastal heath. Ecology 63(5), 1332-1341.
Fox BJ (1995) Eastern chestnut mouse Pseudomys gracilicaudatus. In 'The Mammals of Australia.' 2nd edn. Ed. R Strahan) pp. 601-3. (Reed: Chatswood) Fox BJ, Monamy V (2007) A review of habitat selection by the swamp rat Rattus lutreolus (Rodentia: Muridae). Austral Ecology 32(8), 837-849. Fox BJ, Taylor JE, Thompson PT (2003) Experimental manipulation of habitat structure: a retrogression of the small mammal succession. Journal of Animal Ecology 72(6), 927- 940. Frank PA, Layne JN (1992) Nests and daytime refugia of cotton mice (Peromyscus gossypinus) and golden mice (Ochrotomys nuttalli) in south central Florida. American Midland Naturalist 127(1), 21-30. Friend GR (1993) Impact of fire on small vertebrates in mallee woodlands and heathlands of temperate Australia: a review. Biological Conservation 65(2), 99-114. Gill AM, Ingwersen F (1976) Growth of Xanthorrhoea australis R.Br. in relation to fire. Journal of Applied Ecology 13(1), 195-203. Gray SJ, Hurst JL, Stidworthy R, Smith J, Preston R, MacDougall R (1998) Microhabitat and spatial dispersion of the grassland mouse (Mus spretus Lataste). Journal of Zoology 246, 299-308. Haering R, Fox BJ (1995) Habitat utilization patterns of sympatric populations of Pseudomys gracilicaudatus and Rattus lutreolus in coastal heathland: a multivariate analysis. Australian Journal of Ecology 20(3), 427-441. Haim A, Rozenfeld FM (1995) Temporal segregation in coexisting Acomys species: the possible role of nest site. Journal of Arid Environments 29(4), 505-509. Hall LS, Morrison ML (1997) Den and relocation site characteristics and home ranges of Peromyscus truei in the White Mountains of California. Great Basin Naturalist 57(2), 124-130. Happold M (1976) Social behavior of conilurine rodents (Muridae) of Australia. Zeitschrift Fur Tierpsychologie-Journal of Comparative Ethology 40(2), 113-182.
Pereoglou, F., Population processes in an early successional heathland 40 Chapter 2:Refuge site selection
Hartung TG, Dewsbury DA (1979) Nest-building behavior in seven species of muroid rodents. Behavioral and Neural Biology 27(4), 532-539. Higgs P, Fox BJ (1993) Interspecific competition: a mechanism for rodent succession after fire in wet heathland. Australian Journal of Ecology 18(2), 193-201. Hinkelman TM, Loeb SC (2007) Effect of woody debris abundance on daytime refuge use by cotton mice. Southeastern Naturalist 6(3), 393-406. Ingwersen F (1977) Vegetation development after fire in the Jervis Bay Territory. Masters Thesis, The Australian National University, Canberra Keith DA, McCaw WL, Whelan RJ (2002) Fire regimes in Australian heathlands and their effects on plants and animals. In 'Flammable Australia: The Fire Regimes and Biodiversity of a Continent.' (Eds RA Bradstock, JE Williams and AM Gill) pp. 199-237. (Cambridge University Press: New York) Kemper CM (1981) Description of Pseudomys novaehollandiae burrows located with radioisotopes. Australian Mammalogy 4, 141-3. Klein HG, Layne JN (1978) Nesting behaviour in four species of mice. Journal of Mammalogy 59(1), 103-108. Letnic M (2003) The effects of experimental patch burning and rainfall on small mammals in the Simpson Desert, Queensland. Wildlife Research 30(6), 547-563.
Letnic M, Dickman CR, Tischler MK, Tamayo B, Beh CL (2004) The responses of small mammals and lizards to post-fire succession and rainfall in arid Australia. Journal of Arid Environments 59(1), 85-114. Lindenmayer DB, Cunningham RB, Peakall R (2005) The recovery of populations of bush rat Rattus fuscipes in forest fragments following major population reduction. Journal of Applied Ecology 42(4), 649-658. Lindenmayer DB, MacGregor C, Welsh A, Donnelly C, Crane M, Michael D, Montague-Drake R, Cunningham RB, Brown D, Fortescue M, Dexter N, Hudson M, Gill AM (2008) Contrasting mammal responses to vegetation type and fire. Wildlife Research 35(5), 395- 408. Lindenmayer DB, MacGregor C, Wood JT, Cunningham RB, Crane M, Michael D, Montague- Drake R, Brown D, Fortescue M, Dexter N, Hudson M, Gill AM (2009) What factors influence rapid post-fire site re-occupancy? A case study of the endangered eastern bristlebird in eastern Australia. International Journal of Wildland Fire 18(1), 84-95. Lindenmayer DB, Welsh A, Donnelly CF, Meggs RA (1996) Use of nest trees by the mountain brushtail possum (Trichosurus caninus) (Phalangeridae: Marsupialia) I. Number of occupied trees and frequency of tree use. Wildlife Research 23(3), 343-361.
Pereoglou, F., Population processes in an early successional heathland 41 Chapter 2:Refuge site selection
Luo J, Fox BJ (1994) Diet of the eastern chestnut mouse (Pseudomys gracilicaudatus) II Seasonal and successional patterns. Wildlife Research 21(4), 419-431. Luo J, Fox BJ (1995) Competitive effects of Rattus lutreolus presence on food resource use by Pseudomys gracilicaudatus. Australian Journal of Ecology 20(4), 556-564. Luo J, Fox BJ, Jefferys E (1994) Diet of the eastern chestnut mouse (Pseudomys gracilicaudatus) I Composition, diversity and individual variation. Wildlife Research 21(4), 401-417. Madison DM, Hill JP, Gleason PE (1984) Seasonality in the nesting behavior of Peromyscus leucopus. American Midland Naturalist 112(1), 201-204.
Masters P (1993) The effects of fire driven succession and rainfall on small mammals in spinifex grassland at Uluru National Park, Northern Territory. Wildlife Research 20(6), 803-813. McCullagh P, Nelder JA (1989) 'Generalized linear models.' 2nd edn. (Chapman and Hall) McCulloch CE, Searle SR, Neuhaus JM (2008) 'Generalized, linear, and mixed models.' 2nd edn. (Wiley: Hoboken, N.J. ) Monamy V, Fox BJ (2000) Small mammal succession is determined by vegetation density rather than time elapsed since disturbance. Austral Ecology 25(6), 580-587. Monamy V, Fox BJ (2010) Responses of two species of heathland rodents to habitat manipulation: vegetation density thresholds and the habitat accommodation model. Austral Ecology 35(3), 334-347. Moro D, Morris K (2000) Movements and refugia of Lakeland Downs short-tailed mice, Leggadina lakedownensis, and house mice, Mus domesticus, on Thevenard Island, Western Australia. Wildlife Research 27(1), 11-20. Morris DW, Fox BJ, Luo J, Monamy V (2000) Habitat-dependent competition and the coexistence of Australian heathland rodents. Oikos 91, 294-306. Morzillo AT, Feldhamer GA, Nicholson MC (2003) Home range and nest use of the golden mouse (Ochrotomys nuttalli) in Southern Illinois. Journal of Mammalogy 84(2), 553-560.
Newsome AE, Catling PC (1983) Animal demography in relation to fire and shortage of food: some indicative models. Ecological Studies 43, 490-505. Newsome AE, McIlroy J, Catling PC (1975) The effects of an extensive wildfire on populations of twenty ground vertebrates in southeast Australia. Proceedings of the Ecological Society of Australia 9, 107-123. Ostfeld RS (1985) Limiting resources and territoriality in microtine rodents. American Naturalist 126(1), 1-15.
Pereoglou, F., Population processes in an early successional heathland 42 Chapter 2:Refuge site selection
Sanecki GM, Green K, Wood H, Lindenmayer D, Sanecki KL (2006) The influence of snow cover on home range and activity of the bush-rat (Rattus fuscipes) and the dusky antechinus (Antechinus swainsonii). Wildlife Research 33(6), 489-496. Schwarz G (1978) Estimating the dimension of a model. Annals of Statistics 6, 461-464. Specht RL (1981) Responses to fire in heathlands and related shrublands. In 'Fire and the Australian Biota.' (Eds AM Gill, RH Groves and IR Noble) pp. 395-415. (Australian Academy of Science: Canberra) Stokes MK (1995) Selection of refuge sites by sympatric Microtus ochrogaster and Sigmodon hispidus. Journal of Mammalogy 76(1), 83.
Strahan R (Ed.) (1995) 'The Mammals of Australia (2nd edn).' (Reed: Chatswood) Sutherland EF, Dickman CR (1999) Mechanisms of recovery after fire by rodents in the Australian environment: a review. Wildlife Research 26(4), 405-419. Taws N (1997) Vegetation survey and mapping of Jervis Bay Territory. Report, Environment Australia, Canberra. Taylor JM, Horner BE (1972) Observations on reproductive biology of Pseudomys (Rodentia: Muridae). Journal of Mammalogy 53(2), 318-328. Thompson MB, Medlin G, Hutchinson R, West N (1989) Short-term effects of fuel reduction burning on populations of small terrestrial mammals. Australian Wildlife Research 16(2), 117-129. Thompson PT, Fox BJ (1993) Asymmetric competition in Australian heathland rodents: a reciprocal removal experiment demonstrating the influence of size-class structure. Oikos 67(2), 264-278. Townley S (2007) Targeted survey for the eastern chestnut mouse Pseudomys gracilicaudatus, New Holland mouse Pseudomys novaehollandiae, eastern pygmy-possum Cercartetus nanus, and long-nosed bandicoot Perameles nasuta on Pacific City Estate lands, Jervis Bay National Park. NSW Department of Environment. Townley SJ (2000) The Ecology of the Hastings River Mouse Pseudomys oralis (Rodentia: Muridae) in Northeastern New South Wales and Southeastern Queensland. PhD Thesis, Southern Cross University, Lismore Wagner DM, Feldhamer GA, Newman JA (2000) Microhabitat selection by golden mice (Ochrotomys nuttalli) at arboreal nest sites. American Midland Naturalist 144(1), 220- 225. Ward GD (1978) Habitat use and home range of radio-tagged opossums Trichosurus vulpecula (Kerr) in New Zealand lowland forest. In 'The Ecology of Arboreal Folivores.' Ed. GG Montogomery) pp. 305-313. (Smithsonian Institute Press: Washington DC)
Pereoglou, F., Population processes in an early successional heathland 43 Chapter 2:Refuge site selection
West SD, Dublin HT (1984) Behavioural strategies of small mammals under winter conditions: solitary or social? In 'Proceedings of the International Colloquium: Winter Ecology of Small Mammals.' Ed. JF Merritt) pp. 293-299. (Carnegie Museum of Natural History: Pittsburgh) Whelan RJ (1995) 'The Ecology of Fire.' (Cambridge University Press: Cambridge) Whelan RJ, Rodgerson L, Dickman CR, Sutherland EF (2002) Critical life cycles of plants and animals: developing a process-based understanding of population changes in fire-prone landscapes. In 'Flammable Australia: The Fire Regimes and Biodiversity of a Continent.' (Eds RA Bradstock, JE Williams and AM Gill) pp. 94-124. (Cambridge University Press: New York) Wolfe JL (1970) Experiments on nest-building behaviour in Peromyscus (Rodentia: Cricetinae). Animal Behaviour 18(Part 4), 613-615. Wolff JO, Hurlbutt B (1982) Day refuges of Peromyscus leucopus and Peromyscus maniculatus. Journal of Mammalogy 63(4), 666-668. Wolton RJ, Trowbridge BJ (1985) The ranging and nesting behavior of wood mice, Apodemus sylvaticus (Rodentia: Muridae), as revealed by radio-tracking. Journal of Zoology 206(Jun), 203-224. Woods RE, Ford FD (2000) Observations on the behaviour of the smoky mouse Pseudomys fumeus (Rodentia: Muridae). Australian Mammalogy 22, 35-42.
Pereoglou, F., Population processes in an early successional heathland 44
Chapter 3: Site occupancy
Site occupancy
Published as:
Pereoglou, F., MacGregor, C., Banks, S.C., Ford, F., Wood, J., Lindenmayer, D.B. Landscape, fire and habitat: which features of recently burned heathland influence site occupancy of an early successional specialist? Landscape Ecology 31(2): 255-269. doi:10.1007/s10980-015-0240-2
Manuscript: Appendix 2
Pereoglou, F., Population processes in an early successional heathland 45 Chapter 3: Site occupancy
3.1 Abstract
Multiple ecological drivers generate spatial patterns in species’ distributions. Changes to natural disturbance regimes can place early successional habitat specialists at an increased risk of extinction by altering landscape patterns of habitat suitability. We developed a series of hypotheses to evaluate the effects of landscape structure, fire history, and site-level habitat quality on site occupancy by an early successional specialist, the eastern chestnut mouse (Pseudomys gracilicaudatus). We obtained eight years of monitoring data from 26 sites in recently burned heathland in southeast Australia. We used generalised linear models to determine which explanatory variables were related to occupancy. We also explored predictability in patterns of small mammal species co-occurrence. Landscape structure (patch area, landscape heterogeneity) was strongly related to site occupancy. Site occupancy was associated with dead shrubs in the understory and rock cover on ground layer, but was not directly influenced by recent or historical fire. Contrary to contemporary ecological theory, we found no predictable species associations in our early successional community. We recommend surveys take account of landscape configuration and proximity to suitable habitat for optimal results. Fire regimes expected to promote eastern chestnut mouse population growth should encourage the retention of critical habitat features rather than be based on temporal rates of successional stages. For management to adequately account for post-disturbance patterns in early successional communities, a species-by-species, multi-scaled approach to research is necessary.
3.2 Introduction
Early successional communities are the starting point for recovery after major disturbance. These communities are comprised of residual individuals, colonising opportunists and habitat specialist species (Connell and Slatyer 1977; Swanson et al. 2011). Understanding the ecological processes operating in early successional communities is important, given these ephemeral habitats are confined to the variable length of time between a disturbance, and re- establishment of the later successional stages of community composition (Swanson et al. 2011). Estimating the duration of the early successional community is difficult as it is related to the type of ecosystem and is likely to fluctuate within ecosystems of the same type (Bradstock et al. 2012); for example, years for heathlands (Keith et al. 2002) compared to decades or even centuries in some forests (McCarthy et al. 1999). This time frame is likely to vary geographically (Velle and Vandvik 2014), and be dependent on the type and frequency of disturbance events (e.g. Sousa 1979; Delmoral and Bliss 1993; Rossi et al. 2007; Gill and Allan 2008).
Pereoglou, F., Population processes in an early successional heathland 46 Chapter 3: Site occupancy
Heathland communities often occur in the early successional stage of regeneration after disturbance. In Australia, heathlands are one of the most fire-prone and flammable ecosystems (Williams 1995; Keith et al. 2002). Following disturbance, these heathlands can accumulate sufficient plant biomass for fire ignition every two years (Williams 1995) and reach a mature to senescent phase after 25 to 50 years (Specht 1981). The compositional dynamics of heathland are primarily driven by the fire regime. High fire frequency (i.e. every two years) can reduce a heathland to a sedgeland by eliminating woody shrubs and allowing herbaceous species to dominate (Williams 1995). Long intervals between fires (i.e. greater than 20 years) will produce closed scrubs with diminished understory floristic richness (Williams 1995; Keith et al. 2002). Heathlands generally occur as patches of various sizes within a heterogeneous landscape (Williams 1995). Fire occurs over multiple spatial and temporal scales, with variable impacts on biodiversity and key ecosystem processes that are strongly site- and species-specific (Pickett and White 1985; Turner et al. 1998; Whelan et al. 2002). For small mammals, patterns of population responses to fire have been studied over multiple decades (Newsome et al. 1975; Masters 1993; Letnic et al. 2013). Small mammals are associated with a range of fire frequencies; from high (e.g. heath rat (Pseudomys shortridgei), Cockburn et al. 1981; eastern chestnut mouse (P. gracilicaudatus), New Holland mouse (P. novahollandiae), Fox 1982) to low (e.g. desert mouse (P. desertor), Letnic and Dickman 2005; brown antechinus (Antechinus stuartii), bush rat (Rattus fuscipes), swamp rat (R. lutreolus), Lindenmayer et al. 2008a; silky mouse (P. apodemoides), Di Stefano et al. 2011). Changes to fire regimes can reduce habitat suitability for some species (e.g. Fox 1982; Baker et al. 2010). Contingent upon a species’ location along the continuum, increased fire frequency may improve population viability. For early successional species, fire suppression can lead to population extinction through habitat degradation and impaired demographic processes (Hutto 2008; Templeton et al. 2011). For example, with the re-introduction of fire, Templeton et al. (2011) showed isolated patches could transition into a stable metapopulation with concomitant changes in the amount and patterns of animal movements across the landscape. Furthermore, fire-specialist fauna have been shown to occur with increasing likelihood as spatial proximity to fire, or fire severity increases (Hutto 2008), and have positive population growth rates in habitat created by fire (Rota et al. 2014). For early successional heathland specialists, the local physical conditions that are associated with their presence are likely to be in a continual state of change. The ability to reverse population declines for these species is complicated by the unpredictable, spatially and temporally variable nature of their habitats. However, whether fire plays a positive role in species occurrence and assists in maintaining habitat suitability in early successional communities is poorly understood.
Research across multiple spatial scales is necessary for identifying the processes underlying species’ responses in patchy and temporally-dynamic communities (Levin 1992;
Pereoglou, F., Population processes in an early successional heathland 47 Chapter 3: Site occupancy
Thornton et al. 2011). Most theoretical and empirical research in disturbance ecology has focussed on site-level disturbance history (e.g. time since fire). However, multiple ecological drivers together generate spatial patterns in species’ distributions (Turner 2005). For instance, research in landscape ecology has a strong focus on landscape pattern variables such as patch size and landscape heterogeneity, as well as site-level habitat quality (Burel and Baudry 2003). Here, we tested the effects of landscape patterns on spatial processes by integrating fire history (sensu landscape legacy (Turner 2005) or ecological memory (Peterson 2002)), with classic landscape ecology themes of patch size and landscape heterogeneity, as well as community ecology themes of interspecific interactions. We investigated landscape patterns of habitat suitability across multiple spatial scales using a species described as an early successional specialist, the eastern chestnut mouse (Pseudomys gracilicaudatus, Gould 1845). Specifically, we assessed a series of hypotheses relating to the perceived importance of fire history, landscape, community and site-level habitat features influencing site occupancy in an early successional heathland (Fig. 3.1).
Figure 3.1: Conceptual diagram showing the predicted effects of explanatory variables on eastern chestnut mouse site occupancy; black arrows indicate a positive effect, white arrows indicate a negative effect, grey arrows indicate no effect.
Hypothesis 1 (fire history): We predicted occupancy was positively related to the distribution of the most recent fire, positively related to a high incidence of fires, and negatively related to the years since last fire (Fig. 3.1). Recent fire creates suitable habitat for early successional specialists (Monamy and Fox 2000). Therefore, with either a recent fire or an
Pereoglou, F., Population processes in an early successional heathland 48 Chapter 3: Site occupancy increased incidence of fire at a site, post-fire recruitment by residual survivors should offset the potential for local extinction. Hypothesis 2 (site-level habitat quality): We predicted occupancy was positively related to understory floristic richness, the number of dead shrubs in the understory, and live vegetative cover on the ground layer (Fig. 3.1). The distribution of species after disturbance is dependent on the availability of suitable local habitats, and has been linked to shelter (Fox et al. 2003; Robinson et al. 2013), food (Luo and Fox 1996; Vieira and Briani 2013), and breeding resource requirements (Banks et al. 2005b; Watson et al. 2012). The floristic richness of early successional heathland provides a varied diet for an opportunistic generalist herbivore like the eastern chestnut mouse (Luo et al. 1994). Dense vegetation on the ground layer and a collapsed dead shrub layer are significant components of the species’ diurnal shelter requirements (Pereoglou et al. 2011). Hypothesis 3 (interspecific interactions): We predicted eastern chestnut mouse site occupancy was negatively related to swamp rat (Rattus lutreolus) occurrence, but was not influenced by the occurrence of the bush rat (R. fuscipes) or brown antechinus (Antechinus stuartii) (Fig. 3.1). After a disturbance, species are added to the community assemblage when their specific habitat requirements are met (habitat accommodation model; Fox 1982). Species replacement is a result of subsequent competitive interactions that occur with time since fire or vegetation succession (Fox et al. 2003). The swamp rat is the only one of the co-occurring species that is similar to the eastern chestnut mouse in its resource requirements and has also been shown to be competitively superior (Higgs and Fox 1993). Hypothesis 4 (landscape context): We predicted site occupancy was positively related to heathland patch area and negatively influenced by landscape heterogeneity (Fig. 3.1). The probability of species occurrence in a habitat patch varies as a function of patch size (MacArthur and Wilson 1967) with a species’ area sensitivity being greater when the intervening matrix is inhospitable (Prugh et al. 2008). The eastern chestnut mouse is considered a heathland habitat specialist (Fox 1982) but there has been no study of how habitat availability and heterogeneity influences the species’ local distribution across a landscape.
3.3 Materials and Methods
3.3.1 Study species
The eastern chestnut mouse is an endemic Australian murid rodent. It has a disjunct distribution along the east coast of Australia (inset Fig. 3.2) and is absent throughout much of its range. The species inhabits heathland in southern Australia and has been recorded in woodland in the northern parts of its range (Fox 2008). Usually reported at low density, populations peak in abundance post-fire before declining as vegetation regenerates (Fox 1982). Threats to the
Pereoglou, F., Population processes in an early successional heathland 49 Chapter 3: Site occupancy viability of populations of the species are thought to include fire suppression, loss of heath habitat, and predation (Fox 2008). A summarised description of the eastern chestnut mouse and the co-occurring small mammal species with which it is expected to have competitive interactions is provided (Table 3.1; a full version of species’ descriptions available in Appendix 4).
3.3.2 Study area and survey design
Booderee National Park is a ~6 500 ha reserve located in the Jervis Bay region, approximately 200 km south of Sydney, Australia (Fig. 3.2). Booderee National Park is characterised by natural heterogeneity in vegetation with six broad vegetation types (forest, rainforest, woodland, heathland, shrubland and sedgeland) identified in previous vegetation surveys (Taws 1997) (Fig. 3.2). Monitoring heathland sites (n = 26) began in 2003 as part of a long-term study on the effects of fire on vertebrate biota (see Lindenmayer et al. 2008a). Site monitoring was conducted annually during the summer months for the period 2003 – 2011. Sites comprised a single 100 m transect with: (i) a black plastic drift fence connecting 380 mm deep x 300 mm diameter pitfall traps at 0 and 20 m, 40 and 60 m, and 80 and 100 m; (ii) large wire cage traps (30 x 30 x 60 cm) placed at 0 m and 100 m with four small wire cage traps (20 x 20 x 50 cm) at 20 m intervals between; and (iii) ten aluminium box traps (Elliott Scientific Equipment, Australia) placed every 10 m along the transect line from 0 – 90 m (Fig. 3.3). We baited wire cage traps and aluminium box traps with rolled oats and peanut butter. Effort varied between sites and survey years, but on average, we opened traps for three consecutive nights each survey year. We checked all traps daily during the trapping period and recorded species, sex, age class, trap number and trap type. Animals were marked to identify recaptures within a single survey year trapping period before release at the point of capture. Vegetation was sampled during the survey years of 2004/05, 2006/07, 2007/08, and 2009/10 in two 20 x 20 m plots located at the 20 – 40 m and 60 – 80 m sections of each transect (Fig. 3.3). We counted the number of dead shrubs (where a shrub was defined as single woody plant either multi-stemmed at the base or, if single stemmed, less than 2m tall; dead shrubs, Table 3.2), the number of understory plant species present (floristic richness, Table 3.2), and visually estimated the understory (< 2 m) percent live cover of all plant species combined (cover, Table 3.2). Values were averaged across the two vegetation plots. Percent rock cover, bare ground and regenerating shrubs in the ground layer were visually estimated in four 1 x 1 m sub-plots nested within the larger vegetation plot, values were averaged (Table 3.2). Fire history data was extracted from GIS layers available for Booderee National Park using ArcGIS version 9.2 (ESRI 2009). At each site, for each survey year, we calculated the
Pereoglou, F., Population processes in an early successional heathland 50 Chapter 3: Site occupancy
Table 3.1: Summarized description of the eastern chestnut mouse and co-occurring small mammal species. Full description provided in Appendix S1 in Appendix 4 Species Habitat Average adult Breed: all Litter Breed in Lifespan Diet Habit Source weight (grams) year (A) or size season of (years) season (S) birth Eastern chestnut mouse Woodland 90 (male) S 1 – 5 (3) No < 2 Herbivorous Nocturnal, Fox 1982; Fox and Kemper (Pseudomys gracilicaudatus) Heathland 75 (female) terrestrial, 1982; Luo et al. 1994; Fox territorial 2008 Bush rat Widespread > 200 A, S 1 – 7 (5) Yes 1 – 2 Omnivorous Nocturnal, Warneke 1971; Watts and (Rattus fuscipes) terrestrial, Aslin 1981; Robinson mostly 1987; Breed and Ford territorial 2007; Lunney 2008a Swamp rat Heathland 200 A, S 3 – 5 Yes 1 – 2 Herbivorous Partially Watts and Aslin 1981; (R. lutreolus) Sedgeland diurnal, Braithwaite 1982; Lunney terrestrial, 2008b territorial Brown antechinus Forest 35 (male) S 8 – 10 No 1 (male) Insectivorous Terrestrial or Fox and Archer 1984; (Antechinus stuartii) Heathland 20 (female) 1 – 2 (female) arboreal, Cockburn and Lazenby- communal Cohen 1992; Crowther nesting 2002; Crowther and Braithwaite 2008; Dickman 2008
Pereoglou, F., Population processes in an early successional heathland 51 Chapter 3: Site occupancy
Figure 3.2: Map of Booderee national park indicating study sites and vegetation types. Inset shows the distribution of the eastern chestnut mouse and the study area location.
Figure 3.3: Layout of a site transect and the location of vegetation plots. Diagram is not to scale.
Pereoglou, F., Population processes in an early successional heathland 52 Chapter 3: Site occupancy
number of years elapsed since the last fire (years since fire, Table 3.2), and counted the number of fires affecting the site since fire mapping began in the mid-1950s (previous fires, Table 3.2). Sites were scored with a 0 or 1 to indicate whether they were burned in the most recent major fire in December 2003 (burned 2003, Table 3.2). Eighty percent of heathland sites were burned in the 2003 fire. Landscape structure was measured from GIS layers of the vegetation mapping of Taws (1997) available for Booderee National Park, using ArcGIS version 9.2 (ESRI 2009). As a measure of landscape heterogeneity, we determined the number of vegetation types within a polygon of 500 m radius around each site (landscape heterogeneity, Table 3.2). A 500 m radius was chosen because the closest sites were 1 km apart. We calculated the area of the mapped polygon of heathland within which a site was located (patch area, Table 3.2). Landscape heterogeneity and patch area were fixed for the study period.
3.3.3 Statistical analysis
We used Spearman’s rank correlation (Conover 1999) to identify redundant variables. We explored the individual significance of the explanatory variables (Table 3.2, Fig. 3.1) using a series of univariate hierarchical generalised linear models (HGLMs; Lee et al. 2006) with an over-dispersed Poisson distribution and logarithm-link function. Our model response variable, site occupancy, was the total number of eastern chestnut mouse captures across trap nights at each site. We fitted each explanatory variable individually in separate models that included site as a random term to account for variation among sites, and survey year as a fixed effect (except where site occupancy was predicted by years since the last fire, survey year was fitted as a random term). We calculated effort as the total number of traps open at each site multiplied by the total number of nights that traps were open within a survey year and we included the log of effort as an offset variable to account for unequal survey effort between sites and survey years (Table S1 in Appendix 4). To check for possible interrelationships between the effects of the explanatory variables, we also fitted generalized linear models (GLMs; McCullagh and Nelder 1989) for all possible subsets of the set of potential explanatory variables (Table 3.1, Fig. 3.1). We evaluated the fits using Schwarz information criterion (SIC; Schwarz 1978) and Akaike’s information criterion (AIC; Burnham and Anderson 2002). Complex fixed effects models are computationally much easier to fit using GLMs. Terms that did not contribute anything useful in the HGLMs might be retained, but it was very unlikely we would omit a term that was important. We also looked for an indirect effect of fire via other explanatory variables by testing whether the apparent effect of fire was removed when another explanatory variable was included in our models, and in turn fitting a model to test whether that variable was significantly affected by fire. Based on the explanatory variables identified in the GLM of all subsets and the univariate HGLMs, we then fitted a multivariate HGLM using the predictors from the ‘best’ model. Pereoglou, F., Population processes in an early successional heathland 53 Chapter 3: Site occupancy Finally, we fitted the captures of other small mammals (bush rat, swamp rat and brown antechinus) as dependent variables in a series of separate models to determine whether the eastern chestnut mouse responded differently compared to sympatric species in relation to fire history, landscape structure and site-level habitat quality.
Table 3.2: Description of explanatory variables used to predict eastern chestnut mouse site occupancy
Variable Description Mean S.E. Min Max Survey year 2003 – 2010/11 (n = 9) Mean S.E. Min Max Site Unique site identifier (n = 26) Effort Total number of traps open x total number of nights traps were open 63.30 2.09 0 138 open 63.30 2.09 0 1 Species associations
Eastern chestnut mouse Eastern chestnut mouse captures (new animals + recaptures) 0.63 0.11 0 12 0.63 0.11 0 Bush rat Bush rat captures (new animals + recaptures) 2.78 0.28 0 24 2.78 0.28 0 24 Swamp rat Swamp rat captures (new animals + recaptures) 0.45 0.10 0 12 0.45 0.10 0 Brown antechinus Brown antechinus captures (new animals + recaptures) 1.89 0.16 0 11 1.89 0.16 0 Fire history Burned 2003 Site burned (1) or not (0) in the December 2003 fire Number of previous fires Count of previous fires recorded at each site 4.54 0.120.12 1 1 9 Years since fire Number of years elapsed since the last fire at each site Site-level habitat quality Dead shrub Count of dead shrubsA 28.1028.10 3.253.25 0 0 270270 Floristic richness Number of understory speciesA 13.013.09 0.0.4646 4.54.5 26.5026.50 Percent cover Percent live cover of all understory species combinedA A 85.4485.44 1.1.7979 1515 100100 Percent rock Percent rock cover in ground layerB 1.24 0.0.3939 0 0 24.3824.38 Percent bare Percent cover of bare ground in ground layerB 17.4117.41 2.122.12 0 0 81.8881.88 B Percent regenerating vegetation Percent cover of regenerating vegetation in ground layerB 16.7116.71 2.632.63 0 0 100100
Landscape structure i 88.59 8.74 0.93 362.20 Patch area Number of hectares of contiguous patch of vegetation of type i 88.59 8.74 0.93 362.20 4.00 0.07 2 Landscape heterogeneity Number of vegetation types in 500 m radius of site 4.00 0.07 2 6 AAverage of data measured in two 20 x 20 m plots BAverage of data measured in four 1 x 1 m subplots
3.4 Results
We completed a total of 14812 trap nights and made 1346 captures (including 366 recaptures) of the eastern chestnut mouse and co-occurring terrestrial small mammal species.
3.4.1 Correlation between explanatory variables
We found a low level of correlation between explanatory variables (Table 3.3). Variable pairs with the highest levels of correlation included patch area and landscape heterogeneity, previous fires and landscape heterogeneity, years since last fire and burned 2003, and bush rat and effort. The landscape structure variables represented different effects and were not so highly correlated that inclusion together substantially reduced the reliability of the further modelling. The extent of the 2003 fire meant insufficient fire ages were available to model a time since fire
Pereoglou, F., Population processes in an early successional heathland 54 Chapter 3: Site occupancy effect on occupancy so we removed years since fire from further analysis and retained the binary variable.
3.4.2 Fire history
Although the eastern chestnut mouse was only observed on burned sites, the direct effect of burning in 2003 was not significant. This partly reflects the small number of heathland sites that were not burned in the most recent fire. Whether a site was burned in 2003 had a positive effect on eastern chestnut mouse site occupancy when the percent rock cover on the ground layer was included in the model (Table 3.4). However, all of the sites that were not burned in 2003 had no rock, so the effects are confounded to some extent and it is mainly a percent rock cover effect. Hence, H1 was rejected, as we were unable to detect a direct effect of either previous fires or burning in the 2003 fire on eastern chestnut mouse site occupancy (Table S2 in Appendix 4, Fig. 3.4).
3.4.3 Site-level habitat quality
The number of dead shrubs in the understory had a weak positive effect on eastern chestnut mouse site occupancy (Fig. 3.4). An exponential increase in the number of dead shrubs in the understorey was associated with a small increase in eastern chestnut mouse captures per 100 trap nights (Fig. 3.5a). As an indirect effect of fire, we found the number of dead shrubs significantly increased with burning in 2003. Percent rock cover on the ground layer positively influenced eastern chestnut mouse site occupancy in burned sites (Table 3.4). Understory floristic richness and percent cover did not affect eastern chestnut mouse site occupancy, nor did the percent bare ground or regenerating shrubs in the ground layer. We accepted H2, habitat features of the understory and ground layer influence eastern chestnut mouse site occupancy (Table S2 in Appendix 4, Fig. 3.4).
3.4.4 Species associations
We found a negative effect of bush rat abundance on eastern chestnut mouse site occupancy (Fig. 3.4). There was no significant influence of swamp rat or brown antechinus abundance on eastern chestnut mouse site occupancy. When we included the occurrence of the bush rat, swamp rat and brown antechinus as fixed terms in a model together, no species had a significant influence on eastern chestnut mouse site occupancy (Table S2 in Appendix 4).
Hence, H3 was rejected, as co-occurring small mammal species did not influence eastern chestnut mouse site occupancy as predicted (Table S2 in Appendix 4, Fig. 3.4).
Pereoglou, F., Population processes in an early successional heathland 55 Chapter 3: Site occupancy
Table 3.3: Correlation matrix showing Spearman’s rank correlation coefficients for explanatory variables. Correlated variables ± >0.5 are bold.
Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 1 Effort 1 2 Eastern chestnut mouse -0.10 1 3 Bush rat 0.59 -0.18 1 4 Swamp rat 0.11 -0.06 0.00 1 5 Brown antechinus -0.02 0.03 0.19 0.25 1 6 Burned 2003 -0.02 0.17 -0.50 -0.29 -0.40 1 7 Number of previous fires -0.08 0.02 -0.12 -0.02 0.08 0.18 1 8 Years since fire 0.17 -0.05 0.48 0.01 0.21 -0.55 -0.16 1 9 Dead shrub 0.07 -0.06 0.04 -0.13 -0.10 0.21 0.05 -0.14 1 10 Floristic richness 0.15 0.17 -0.18 0.01 -0.18 0.15 0.05 0.11 -0.09 1 11 Percent cover -0.06 0.11 -0.08 0.17 0.09 0.08 0.11 -0.19 -0.07 0.13 1 12 Percent bare 0.39 -0.17 0.12 -0.18 -0.34 0.25 -0.06 0.02 0.32 0.06 -0.47 1 13 Percent rock -0.02 0.19 -0.16 -0.09 -0.08 0.11 0.08 -0.12 0.05 -0.03 -0.34 0.16 1 14 Percent regenerating vegetation -0.05 0.06 -0.15 -0.04 -0.02 0.10 0.19 -0.29 0.23 0.03 0.08 -0.12 0.21 1 15 Patch area 0.04 0.19 -0.12 -0.14 -0.09 0.22 0.45 -0.13 0.13 0.32 0.04 0.11 0.40 0.49 1 16 Landscape heterogeneity -0.01 -0.03 0.03 -0.19 -0.03 0.10 -0.52 -0.02 0.13 -0.27 -0.22 0.08 -0.13 -0.32 -0.67 1
Pereoglou, F., Population processes in an early successional heathland 56 Chapter 3: Site occupancy
Table 3.4: Ten best generalized linear models from fitting all possible subsets of explanatory variables. ΔSIC and ΔAIC are the differences in SIC and AIC for the given models from the value for the optimal model.
ΔSIC ΔAIC Model terms 0 1.22 Survey year + percent rock + burned 2003 1.37 0 Dead shrub + survey year + percent rock + burned 2003 2.44 1.07 Survey year + percent regenerating vegetation + percent rock + burned 2003 3.17 1.80 Survey year + percent rock + percent bare + burned 2003 3.84 2.47 Survey year + percent cover + percent rock + burned 2003 4.09 2.72 Survey year + floristic richness + percent rock + burned 2003 4.16 2.79 Survey year + percent rock + brown antechinus + burned 2003 4.40 3.03 Survey year + percent rock + patch area + burned 2003 4.41 3.04 Survey year + percent rock + bush rat + burned 2003 4.45 3.08 Survey year + percent rock + burned 2003 + landscape heterogeneity
Figure 3.4: Conceptual diagram showing the modelled effects of explanatory variables on eastern chestnut mouse site occupancy; black arrows indicate a positive effect, white arrows indicate a negative effect, broken line represents indirect effect.
We found direct effects of fire history and landscape structure for the sympatric small mammal species. Burning in the 2003 fire had a strong negative effect on both swamp rat and brown antechinus site occupancy estimates (Table S2 in Appendix 4). A weak negative effect of patch area on brown antechinus and swamp rat site occupancy indicates a potential sensitivity to landscape heterogeneity (Table S2 in Appendix 4).
Pereoglou, F., Population processes in an early successional heathland 57 Chapter 3: Site occupancy 3.4.5 Landscape structure
Landscape heterogeneity had a strong negative effect on eastern chestnut mouse site occupancy (Fig. 3.4), with the number of captures per 100 trap nights decreasing significantly as habitat diversity increased (Fig. 3.5b). To a lesser extent, eastern chestnut mouse site occupancy was affected by patch area (Fig. 3.5c). Hence, we accepted H4, landscape structure affected eastern chestnut mouse site occupancy (Table S2 in Appendix 4, Fig 3.4).
Figure 3.5: Number of eastern chestnut mouse captures predicted from hierarchical generalised linear models over a range of values of significant explanatory variables; (a) number of dead shrubs, (b) patch area, and (c) landscape heterogeneity.
Pereoglou, F., Population processes in an early successional heathland 58 Chapter 3: Site occupancy 3.4.6 ‘Best’ model for predicting eastern chestnut mouse site occupancy
When significant explanatory variables determined from the univariate HGLMs and all subsets regression were included in a single model, burned 2003 and percent rock cover were the only variables whose coefficients were not significantly different from zero (Table 3.5).
Table 3.5: Multivariate HGLM of significant explanatory variables. Model included eastern chestnut mouse site occupancy as response variable, site as a random effect, survey year as a fixed effect, and the log of effort as an offset variable.
Effect Estimate S.E. t P Bush rat -0.116 0.051 -2.26 0.024 Burned 2003 1.035 0.653 1.59 0.113 Percent rock 0.022 0.031 0.72 0.471 Dead shrub 0.010 0.005 2.15 0.032 Patch area 0.004 0.001 2.86 0.004 Landscape heterogeneity -0.354 0.179 -1.98 0.048
3.5 Discussion
3.5.1 Importance of landscape structure versus site-level habitat selection
Our study tested the importance of key themes of ecological theory from disturbance ecology (site-level disturbance history), landscape ecology (site-level habitat quality and landscape patterns) and community ecology (interspecific interactions) on the distribution of an early successional species in a fire-prone environment. A key finding from our study was that site occupancy was strongly influenced by landscape (i.e. broad-scale landscape structure, hypothesis 4), and to a lesser extent, elements within the landscape (i.e. resource availability at a site-level, hypothesis 2) (Fig. 3.4). Large heathland habitat patches, a less heterogeneous intervening habitat matrix, and the presence of dead shrubs indicated a greater likelihood of sites being occupied by the eastern chestnut mouse. This reflects the species’ known shelter resource requirements and presumed habitat specialisation. The basic principle of landscape ecology is that the spatial organization of the environment controls ecological processes (Burel and Baudry 2003). Species respond to patch-level variables because they are important predictors of distribution and abundance (Prugh et al. 2008; Thornton et al. 2011). Habitat specialists that inhabit patchy terrestrial landscapes could be expected to occur predominantly in large habitat patches of the preferred vegetation type and this was the case in our study. Large habitat patches can support higher rates of occupancy compared to smaller ones for species in disturbed landscapes (Thomas et al. 1992; Banks et al.
Pereoglou, F., Population processes in an early successional heathland 59 Chapter 3: Site occupancy
2005a). For early successional species, large habitat patches may facilitate post-fire population recovery as a result of both population expansion from within the fire boundary (Banks et al. 2011) and colonisation of individuals from outside the fire boundary (Watson et al. 2012). Properties of the matrix surrounding habitat patches are also important (Prugh et al. 2008). The surrounding landscape influences patch dynamics and the networks of exchange that exist between habitat patches, facilitating the processes of recolonization (Pickett and White 1985; Burel and Baudry 2003). Therefore, species in isolated patches should have a lower probability of persistence (Turner 1989). In our study, patches were less likely to be occupied as landscape heterogeneity increased. Movement through peripheral habitat can result in direct mortality through predation, or indirectly influence survival through increased competition for sub- optimal resources (Higgs and Fox 1993; Chesson 2000). As other vegetation communities replace heathland, or late seral stage heath replaces early successional communities, species that track suitable environmental conditions may become locally extinct (Templeton et al. 2011). For example, to avoid eastern chestnut mouse extinction, early successional heathland would need to become available close to an extant local population or alternatively, fire itself would need to improve conditions suitable for dispersal (e.g. Templeton et al. 2011; Banks et al. 2013). Analysis of eastern chestnut mouse genetic data supports the latter proposal that recently burned vegetation has greater conductance for gene flow than unburnt habitat (Pereoglou et al. 2013). The interaction between dispersal processes and landscape pattern influences the temporal dynamics of populations (Turner 1989). The capacity of individuals of a population to leave one patch to colonize another similar patch is an essential process of landscape dynamics after disturbance (Burel and Baudry 2003). Conservation must emphasise management of existing habitats and creation of new habitats within colonising distance (Thomas 1994). Species distribution and abundance patterns are influenced by fine-scale features of the environment with which they interact most directly (Thornton et al. 2011). Within-patch habitat attributes such as dead shrubs were important predictors of site occupancy and were positively associated with the most recent major wildfire (Fig. 3.4). This demonstrates that site occupancy of early successional specialists can be facilitated indirectly through the role of fire in creating or maintaining critical habitat features and is indicative of the eastern chestnut mouse preference for sheltering in the collapsed dead shrub layer in recently burned heath (Pereoglou et al. 2011). Habitat features expected to provide shelter have been correlated with animal density for other Pseudomys species – e.g. Xanthorrhea skirt for the heath rat P. shortridgei (Di Stefano et al. 2011), presence of Banksia ornata shrubs for the silky mouse P. apodemoides (Cockburn 1981), and rock cover for the Hastings river mouse P. oralis (Townley 2000). Rock cover was also identified in our modelling as an influence on eastern chestnut mouse occupancy. However, unlike the Hastings river mouse, rock crevasses have not previously been identified as a shelter resource or a preferred microhabitat, so the reasons for this result remain unclear. The variable
Pereoglou, F., Population processes in an early successional heathland 60 Chapter 3: Site occupancy rock cover may be confounded with fire history because sites that were not burnt in the 2003 fire had no rock cover. The level of floristic and structural complexity in regenerating communities has been linked to small mammal species assemblages, and underpins assumptions on species occurrence in successional ecosystems (Catling 1991; Masters 1993; Monamy and Fox 2010). In our study, floristic richness as well as the amount and type of cover were not significant predictors of eastern chestnut mouse site occupancy (Fig. 3.4). Our result is consistent with other investigations, which have uncovered strong species- and site-specific variation in habitat preferences (see also Cockburn 1978; Di Stefano et al. 2011).
3.5.2 Fire history as a predictor of site occupancy for an early successional specialist
Natural disturbances leave legacies that persist for decades to centuries and integrating this history with current understanding of landscape patterns and processes remains an important goal of landscape ecology (Turner 2005). A key finding of this study was that eastern chestnut mouse site occupancy was not directly affected by fire history (hypothesis 1) (Fig. 3.4). Potential explanations for an absence of direct fire history effects are two-fold. First, by comparison with the common co-occurring small mammal species which showed negative responses to fire, the ability to endure disturbance seems to be a necessary trait amongst habitat specialists (Hutto 2008). In another study, where researchers experimentally simulated the effects of fire, populations of the eastern chestnut mouse were also unaffected (Fox et al. 2003). Individuals of the eastern chestnut mouse that survive after fire could be considered ‘biological legacies’ (sensu Franklin et al. 2000). In another investigation in our study region, eastern bristlebirds (Dasyornis brachypterus) that persisted in burned sites also were considered to be survivors of a pre-fire fauna (Lindenmayer et al. 2009). Secondly, the effects of fire history were either captured by, or confounded with other variables. Eighty percent of heathland sites were burned in the 2003 fire, and it is possible that fire history and vegetation type were confounding one another in our analyses. Research that replicates equal numbers of burned and unburned heathland sites may show a different result. We have shown fire to indirectly affect eastern chestnut mouse site occupancy by creating critical habitat features used for diurnal refugia. Therefore, the effects of fire may have been captured entirely by the variable dead shrub. Finally, we may simply not have measured enough fire attributes to model in our analyses. For example, fire regimes are possibly more important than fire history in predicting species occurrence and we were unable to distinguish the relative effects of fire frequency, season and severity (Whelan 1995; Keith et al. 2002).
Pereoglou, F., Population processes in an early successional heathland 61 Chapter 3: Site occupancy 3.5.3 Species interactions and the habitat accommodation model
The habitat accommodation model (Fox 1982) has been shown to be a poor predictor of ecological patterns in relation to fire in other systems (Letnic et al. 2004; Driscoll and Henderson 2008; Di Stefano et al. 2011). The absence of predictable sympatric small mammal species effects on eastern chestnut mouse site occupancy (Fig. 3.4) suggested competitive exclusion was not the strongest biological mechanism structuring the small mammal community as predicted in a mammalian secondary succession (Fox 1982). There was some indication that site occupancy of the eastern chestnut mouse increased in the absence of the bush rat but this effect disappeared when all co-occurring species were modelled together. In other studies, the departure from theoretical predictions has been attributed to various processes including the absence of competitors (Catling 1986; Di Stefano et al. 2011), rapid recovery of vegetation (Lindenmayer et al. 2008b; Lindenmayer et al. 2009) and correlation with alternative explanatory variables (Letnic et al. 2004; Driscoll and Henderson 2008). However, in situ survival of individuals during a fire event probably has a greater influence on population persistence after fire (Lindenmayer et al. 2005; Banks et al. 2011) compared to relative thresholds of habitat suitability. Therefore, it seems unrealistic to expect theoretical models to accurately forecast species assemblages, as they are likely to vary with the complexities of disturbance regimes, spatial patterns of landscape continuity, site-level habitat features, and the interactions among these factors.
3.5.4 Conclusion
Our study evaluates the effects of disturbance, spatial patterns of landscape structure and site-level habitat features on species occupancy in an early successional community. We found a positive effect of heathland patch area and a negative effect of landscape heterogeneity on site occupancy by the eastern chestnut mouse. Eastern chestnut mouse site occupancy also was associated with habitat features, such as a collapsed dead shrub layer, that may be created by burning. We found recent fire and the effect of multiple fires per se did not have a strong influence on eastern chestnut mouse site occupancy. We recommend surveys targeting the species take account of landscape configuration for optimal results. Furthermore, fire regimes expected to promote eastern chestnut mouse population growth should encourage the retention of critical habitat features rather than be based entirely on temporal rates of successional stages. We confirmed the general importance of the multi-level approach to landscape ecology studies, as our study species responded to landscape, patch, and within-patch variables (Thornton et al. 2011). Our work suggests that a multi-scaled approach to investigations of site occupancy should be taken to ensure species- and site-specific resource requirements are identified and that the impacts of changes to disturbance regimes on these resources can be effectively evaluated.
Pereoglou, F., Population processes in an early successional heathland 62 Chapter 3: Site occupancy 3.6 Acknowledgement
We most gratefully acknowledge the Wreck Bay Aboriginal Community and the Wandandian People of the Dharawal and Dhurga language groups on whose lands this study was conducted. Mason Crane, Damien Michael, Rebecca Montague-Drake, and Sachiko Okada provided fieldwork assistance. This project is part of a major study in Booderee National Park funded by the Australian Research Council, the Department of Environment and Water Resources and the Department of Defence.
3.7 References Baker, J., Whelan, R. J., Evans, L., Moore, S. & Norton, M. (2010) Managing the ground parrot in its fiery habitat in south-eastern Australia. Emu, 110, 279-284. Banks, S. C., Cary, G. J., Smith, A. L., Driscoll, D. A., Lindenmayer, D. B., Davies, I. D., Gill, A. M. & Peakall, R. (2013) How does ecological disturbance influence genetic diversity? Trends in Ecology & Evolution, 28, 670-679. Banks, S. C., Dujardin, M., McBurney, L., Blair, D., Barker, M. & Lindenmayer, D. B. (2011) Starting points for small mammal population recovery after wildfire: recolonisation or residual populations? Oikos, 120, 26-37. Banks, S. C., Finlayson, G. R., Lawson, S. J., Lindenmayer, D. B., Paetkau, D., Ward, S. J. & Taylor, A. C. (2005a) The effects of habitat fragmentation due to forestry plantation establishment on the demography and genetic variation of a marsupial carnivore, Antechinus agilis. Biological Conservation, 122, 581-597. Banks, S. C., Ward, S. J., Lindenmayer, D. B., Finlayson, G. R., Lawson, S. J. & Taylor, A. C. (2005b) The effects of habitat fragmentation on the social kin structure and mating system of the agile antechinus, Antechinus agilis. Molecular Ecology, 14, 1789-1801. Bradstock, R. A., Williams, J. & Gill, A. M. (2012) Flammable Australia: The Fire Regimes and Biodiversity of a Continent. Cambridge University Press, Cambridge. Braithwaite, R. W. (1982) Small marginal groups of Rattus lutreolus in good and poor habitat. Journal of Zoology, 198, 529-532.
Breed, B. & Ford, F. (2007) Native Mice and Rats, CSIRO Publishing, Collingwood. Burel, F. & Baudry, J. (2003) Landscape ecology: concepts, methods and applications, Science Publishers, Enfield, USA. Burnham, K. P. & Anderson, D. R. (2002) Model selection and multi-modal inference: a practical information-theoretic approach., Springer, New York. Catling, P. C. (1986) Rattus lutreolus, colonizer of heathland after fire in the absence of Pseudomys species. Australian Wildlife Research, 13, 127-139.
Pereoglou, F., Population processes in an early successional heathland 63 Chapter 3: Site occupancy
Catling, P. C. (1991) Ecological effects of prescribed burning practices on the mammals of southeastern Australia. Conservation of Australia's Forest Fauna (ed D. Lunney), pp. 353-363. Royal Zoological Society of NSW, NSW. Chesson, P. (2000) Mechanisms of maintenance of species diversity. Annual Review of Ecology and Systematics, 31, 343-366. Cockburn, A. (1978) Distribution of Pseudomys shortridgei (Muridae: Rodentia) and its relevance to that of other heathland Pseudomys. Australian Wildlife Research, 5, 213- 219. Cockburn, A. (1981) Diet and habitat preference of the silky desert mouse Pseudomys apodemoides (Rodentia). Australian Wildlife Research, 8, 475-97. Cockburn, A., Braithwaite, R. W. & Lee, A. K. (1981) The response of the heath rat, Pseudomys shortridgei, to pyric succession - a temporally dynamic life-history strategy. Journal of Animal Ecology, 50, 649-666. Cockburn, A. & Lazenby-Cohen, K. A. (1992) Use of nest trees by Antechinus stuartii, a semelparous lekking marsupial. Journal of Zoology, London, 226, 657-680. Connell, J. H. & Slatyer, R. O. (1977) Mechanisms of succession in natural communities and their role in community stability and organization. American Naturalist, 111, 1119- 1144.
Conover, W. J. (1999) Practical Nonparametric Statistics, 3rd Edition, Wiley, New York. Crowther, M. S. (2002) Distributions of species of the Antechinus stuartii-A. flavipes complex as predicted by bioclimatic modelling. Australian Journal of Zoology, 50, 77-91. Crowther, M. S. & Braithwaite, R. W. (2008) Brown antechinus, Antechinus stuartii. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 94-96. Reed New Holland, Sydney, Australia. Delmoral, R. & Bliss, L. C. (1993) Mechanisms of primary succession: insights resulting from the eruption of Mount St Helens. Advances in Ecological Research, 24, 1-66. Di Stefano, J., Owen, L., Morris, R., Duff, T. & York, A. (2011) Fire, landscape change and models of small mammal habitat suitability at multiple spatial scales. Austral Ecology, 36, 638-649. Dickman, C. R. (2008) Agile antechinus, Antechinus agilis. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 83-84. Reed New Holland, Sydney, Australia. Driscoll, D. A. & Henderson, M. K. (2008) How many common reptile species are fire specialists? A replicated natural experiment highlights the predictive weakness of a fire succession model. Biological Conservation, 141, 460-471. ESRI (2009) ArcMap 9.2. Environmental Systems Resource Institute, California.
Pereoglou, F., Population processes in an early successional heathland 64 Chapter 3: Site occupancy
Fox, B. J. (1982) Fire and mammalian secondary succession in an Australian coastal heath. Ecology, 63, 1332-1341. Fox, B. J. (2008) Eastern chestnut mouse, Pseudomys gracilicaudatus. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 634-635. Reed New Holland, Sydney, Australia. Fox, B. J. & Archer, E. (1984) The diets of Sminthopsis murina and Antechinus stuartii (Marsupialia: Dasyuridae) in sympatry. Australian Wildlife Research, 11, 235-248. Fox, B. J. & Kemper, C. M. (1982) Growth and development of Pseudomys gracilicaudatus (Rodentia: Muridae) in the laboratory. Australian Journal of Zoology, 30, 175-185.
Fox, B. J., Taylor, J. E. & Thompson, P. T. (2003) Experimental manipulation of habitat structure: a retrogression of the small mammal succession. Journal of Animal Ecology, 72, 927-940. Franklin, J. F., Lindenmayer, D., MacMahon, J. A., McKee, A., Magnuson, J., Perry, D. A., Waide, R. & Foster, D. (2000) Threads of continuity. Conservation in Practice, 1, 8-17. Gill, A. M. & Allan, G. (2008) Large fires, fire effects and the fire-regime concept. International Journal of Wildland Fire, 17, 688-695. Higgs, P. & Fox, B. J. (1993) Interspecific competition: a mechanism for rodent succession after fire in wet heathland. Australian Journal of Ecology, 18, 193-201.
Hutto, R. L. (2008) The ecological importance of severe wildfires: some like it hot. Ecological Applications, 18, 1827-1834. Keith, D. A., McCaw, W. L. & Whelan, R. J. (2002) Fire regimes in Australian heathlands and their effects on plants and animals. Flammable Australia: the fire regimes and biodiversity of a continent (eds R. A. Bradstock, J. E. Williams & A. M. Gill), pp. 199- 237. Cambridge University Press, New York. Lee, Y., Nelder, J. A. & Pawitan, Y. (2006) Generalized linear models with random effects: unified analysis via h-likelihood, Chapman & Hall/CRC, Boca Raton, FL. Letnic, M. & Dickman, C. R. (2005) The responses of small mammals to patches regenerating after fire and rainfall in the Simpson Desert, central Australia. Austral Ecology, 30, 24- 39. Letnic, M., Dickman, C. R., Tischler, M. K., Tamayo, B. & Beh, C. L. (2004) The responses of small mammals and lizards to post-fire succession and rainfall in arid Australia. Journal of Arid Environments, 59, 85-114. Letnic, M., Tischler, M. & Gordon, C. (2013) Desert small mammal responses to wildfire and predation in the aftermath of a La Nina driven resource pulse. Austral Ecology, In press.
Pereoglou, F., Population processes in an early successional heathland 65 Chapter 3: Site occupancy
Levin, S., A. . (1992) The problem of pattern and scale in ecology: the Robert H. MacArthur award lecture. Ecology, 73, 1943-1967. Lindenmayer, D. B., Cunningham, R. B. & Peakall, R. (2005) The recovery of populations of bush rat Rattus fuscipes in forest fragments following major population reduction. Journal of Applied Ecology, 42, 649-658. Lindenmayer, D. B., MacGregor, C., Welsh, A., Donnelly, C., Crane, M., Michael, D., Montague-Drake, R., Cunningham, R. B., Brown, D., Fortescue, M., Dexter, N., Hudson, M. & Gill, A. M. (2008a) Contrasting mammal responses to vegetation type and fire. Wildlife Research, 35, 395-408.
Lindenmayer, D. B., MacGregor, C., Wood, J. T., Cunningham, R. B., Crane, M., Michael, D., Montague-Drake, R., Brown, D., Fortescue, M., Dexter, N., Hudson, M. & Gill, A. M. (2009) What factors influence rapid post-fire site re-occupancy? A case study of the endangered eastern bristlebird in eastern Australia. International Journal of Wildland Fire, 18, 84-95. Lindenmayer, D. B., Wood, J. T., MacGregor, C., Michael, D. R., Cunningham, R. B., Crane, M., Montague-Drake, R., Brown, D., Muntz, R. & Driscoll, D. A. (2008b) How predictable are reptile responses to wildfire? Oikos, 117, 1086-1097. Lunney, D. (2008a) Bush rat, Rattus fuscipes. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 685-687. Reed New Holland, Sydney, Australia. Lunney, D. (2008b) Swamp rat, Rattus lutreolus. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 690-692. Reed New Holland, Sydney, Australia. Luo, J., Fox, B. J. & Jefferys, E. (1994) Diet of the eastern chestnut mouse (Pseudomys gracilicaudatus) I Composition, diversity and individual variation. Wildlife Research, 21, 401-417. Luo, J. I. A. & Fox, B. J. (1996) Seasonal and successional dietary shifts of two sympatric rodents in coastal heathland: A possible mechanism for coexistence. Austral Ecology, 21, 121-132.
MacArthur, R. H. & Wilson, E. O. (1967) The Theory of Island Biogeography, Princeton University Press, New Jersey. Masters, P. (1993) The effects of fire driven succession and rainfall on small mammals in spinifex grassland at Uluru National Park, Northern Territory. Wildlife Research, 20, 803-813. McCarthy, M. A., Gill, A. M. & Lindenmayer, D. B. (1999) Fire regimes in mountain ash forest: evidence from forest age structure, extinction models and wildlife habitat. Forest Ecology and Management, 124, 193-203.
Pereoglou, F., Population processes in an early successional heathland 66 Chapter 3: Site occupancy
McCullagh, P. & Nelder, J. A. (1989) Generalized linear models, Chapman and Hall, New York. Monamy, V. & Fox, B. J. (2000) Small mammal succession is determined by vegetation density rather than time elapsed since disturbance. Austral Ecology, 25, 580-587. Monamy, V. & Fox, B. J. (2010) Responses of two species of heathland rodents to habitat manipulation: vegetation density thresholds and the habitat accommodation model. Austral Ecology, 35, 334-347. Newsome, A. E., McIlroy, J. & Catling, P. C. (1975) The effects of an extensive wildfire on populations of twenty ground vertebrates in southeast Australia. Proceedings of the Ecological Society of Australia, 9, 107-123. Pereoglou, F., Lindenmayer, D. B., MacGregor, C., Ford, F., Wood, J. & Banks, S. C. (2013) Landscape genetics of an early successional specialist in a disturbance-prone environment. Molecular Ecology, 22, 1267-1281. Pereoglou, F., MacGregor, C., Banks, S. C., Ford, F. D., Wood, J. T. & Lindenmayer, D. B. (2011) Refuge site selection by the eastern chestnut mouse in recently burned heath. Wildlife Research, 38, 290-298. Peterson, G. D. (2002) Contagious disturbance, ecological memory, and the emergence of landscape pattern. Ecosystems, 5, 329-338.
Pickett, S. T. A. & White, P. S. (1985) The Ecology of Natural Disturbance and Patch Dynamics. Academic Press Inc, Florida. Prugh, L. R., Hodges, K. E., Sinclair, A. R. E. & Brashares, J. S. (2008) Effect of habitat area and isolation on fragmented animal populations. Proceedings of the National Academy of Sciences, 105, 20770-20775. Robinson, A. C. (1987) The ecology of the bush rat, Rattus fuscipes (Rodentia: Muridae), in Sherbrooke Forest, Victoria. Australian Mammalogy, 11, 35-49. Robinson, N. M., Leonard, S. W. J., Ritchie, E. G., Bassett, M., Chia, E. K., Buckingham, S., Gibb, H., Bennett, A. F. & Clarke, M. F. (2013) Refuges for fauna in fire-prone landscapes: their ecological function and importance. Journal of Applied Ecology, 50, 1321-1329. Rossi, F., Forster, R. M., Montserrat, F., Ponti, M., Terlizzi, A., Ysebaert, T. & Middelburg, J. J. (2007) Human trampling as short-term disturbance on intertidal mudflats: effects on macrofauna biodiversity and population dynamics of bivalves. Marine Biology, 151, 2077-2090. Rota, C. T., Millspaugh, J. J., Rumble, M. A., Lehman, C. P. & Kesler, D. C. (2014) The role of wildfire, prescribed fire, and mountain pine beetle infestations on the population
Pereoglou, F., Population processes in an early successional heathland 67 Chapter 3: Site occupancy
dynamics of black-backed woodpeckers in the Black Hills, South Dakota. PLoS ONE, 9, e94700. Schwarz, G. (1978) Estimating the dimension of a model. Annals of Statistics, 6, 461-464. Sousa, W. P. (1979) Disturbance in intertidal boulder fields: the non-equilibrium maintenance of species diversity. Ecology, 60, 1225-1239. Specht, R. L. (1981) Responses to fire in heathlands and related shrublands. Fire and the Australian Biota (eds A. M. Gill, R. H. Groves & I. R. Noble), pp. 395-415. Australian Academy of Science, Canberra. Swanson, M. E., Franklin, J. F., Beschta, R. L., Crisafulli, C. M., DellaSala, D. A., Hutto, R. L., Lindenmayer, D. B. & Swanson, F. J. (2011) The forgotten stage of forest succession: early-successional ecosystems on forest sites. Frontiers in Ecology and the Environment, 9, 117-125. Taws, N. (1997) Vegetation survey and mapping of Jervis Bay Territory. Report to Environment Australia. Canberra. Templeton, A. R., Brazeal, H. & Neuwald, J. L. (2011) The transition from isolated patches to a metapopulation in the eastern collared lizard in response to prescribed fires. Ecology, 92, 1736-1747. Thomas, C. D. (1994) Extinction, colonization, and metapopulations: environmental tracking by rare species. Conservation Biology, 8, 373-378. Thomas, C. D., Thomas, J. A. & Warren, M. S. (1992) Distributions of occupied and vacant butterfly habitats in fragmented landscapes. Oecologia, 92, 563-567. Thornton, D. H., Branch, L. C. & Sunquist, M. E. (2011) The influence of landscape, patch, and with-in patch factors on species presence and abundance: a review of focal patch studies. Landscape Ecology, 26, 7-18. Townley, S. J. (2000) The ecology of the Hastings River mouse Pseudomys oralis (Rodentia: Muridae) in northeastern New South Wales and southeastern Queensland. Southern Cross University, Lismore, New South Wales, Australia.
Turner, M. G. (1989) Landscape ecology: the effect of pattern on process. Annual Review of Ecology and Systematics, 20, 171-197. Turner, M. G. (2005) Landscape ecology: what is the state of the science. Annual Review of Ecology and Systematics, 36, 319-344. Turner, M. G., Baker, W. L., Peterson, C. J. & Peet, R. K. (1998) Factors influencing succession: lessons from large, infrequent natural disturbances. Ecosystems, 1, 511-523. Velle, L. G. & Vandvik, V. (2014) Succession after prescribed burning in coastal Calluna heathlands along a 340-km latitudinal gradient. Journal of Vegetative Science, 25, 546- 558.
Pereoglou, F., Population processes in an early successional heathland 68 Chapter 3: Site occupancy
Vieira, E. M. & Briani, D. C. (2013) Short-term effects of fire on small rodents in the Brazillian Cerrad and their relation with feeding habits. International Journal of Wildland Fire, In press. Warneke, R. M. (1971) Field study of the bush rat (Rattus fuscipes). Wildlife Contributions Victoria, 14, 1-115. Watson, S. J., Taylor, R. S., Nimmo, D. G., Kelly, L. T., Clarke, M. F. & Bennett, A. F. (2012) The influence of unburnt patches and distance from refuges on post-fire bird communities. Animal Conservation, 15, 499-507. Watts, C. H. S. & Aslin, H. J. (1981) The Rodents of Australia, Angus & Robertson Publishers, Sydney. Whelan, R. J. (1995) The Ecology of Fire, Cambridge University Press, Cambridge. Whelan, R. J., Rodgerson, L., Dickman, C. R. & Sutherland, E. F. (2002) Critical life cycles of plants and animals: developing a process-based understanding of population changes in fire-prone landscapes. Flammable Australia: The Fire Regimes and Biodiversity of a Continent (eds R. A. Bradstock, J. E. Williams & A. M. Gill), pp. 94-124. Cambridge University Press, New York. Williams, D. G. (1995) Heaths and scrubs. Jervis Bay: A Place of Cultural, Scientific and Educational Value. Kowari 5 (eds G. Cho, A. Georges, R. Stoutjesdijk & R. Longmore), pp. 83-90. Australian Nature Conservation Agency, Canberra.
Pereoglou, F., Population processes in an early successional heathland 69
Chapter 4: Landscape genetics
Landscape genetics
Published as:
Pereoglou, F., Lindenmayer, D.B., MacGregor, C., Ford, F., Wood, J., Banks, S.C. (2013) Landscape genetics of an early successional specialist in a disturbance-prone environment Molecular Ecology 22(5): 1267-1281. doi:10.1111/mec.12172
Manuscript: Appendix 3
Pereoglou, F., Population processes in an early successional heathland 70 Chapter 4: Landscape genetics
4.1 Abstract Species that specialise in disturbed habitats may have considerably different dispersal strategies than those adapted to more stable environments. However, little is known of the dispersal patterns and population structure of such species. This information is important for conservation because many post-fire specialists are at risk from anthropogenic changes to natural disturbance regimes. We used microsatellite markers to assess the effect of landscape variation and recent disturbance history on dispersal by a small mammal species that occupies the early seral stage of vegetation regeneration in burnt environments. We predicted that a post- fire specialist would be able to disperse over multiple habitat types (generalist) and not exhibit sex-biased dispersal; as such strategies should enable effective colonisation of spatially and temporally variable habitat. We found significant differentiation between sites that fitted an isolation-by-distance pattern, and spatial autocorrelation of multilocus genotypes to a distance of 2–3 km. There was no consistent genetic evidence for sex-biased dispersal. We tested the influence of different habitat and fire-specific landscape resistance scenarios on genetic distance between individuals and found a significant effect of fire. Our genetic data supported recently burned vegetation having greater conductance for gene flow than unburnt habitat, but variation in habitat quality between vegetation types and occupied patches had no effect on gene flow. Post-fire specialists must evolve an effective dispersal ability to move over distances that would ensure access to early successional stage vegetation. Natural disturbance and natural heterogeneity may therefore not influence population genetic structure as negatively as expected.
4.2 Introduction
Dispersal behaviour and landscape configuration are primary determinants of population genetic structure (Bowler & Benton 2005; Baguette & Van Dyck 2007). The landscape matrix can facilitate or impede movement of organisms between populations (Taylor et al. 1993; Lindenmayer & Fischer 2006; Fahrig 2007) with dispersal behaviour influencing spatial patterns in abundance (Fahrig & Paloheimo 1988) and habitat selection (Morris 1992). Recently, landscape genetics research has contributed to our understanding of population connectivity in relation to landscape elements (Manel et al. 2003; Storfer et al. 2010). However, threatened species with habitat preferences for disturbed environments are not well represented in the population genetics literature. The ecological requirements of these specialist species suggest that their patterns of dispersal and demography differ considerably from species that utilise more stable environments and that this is likely to be revealed in patterns of genetic diversity and structure. In conjunction with landscape data, spatial genetic information can identify habitat variables that maintain connectivity among populations or serve as barriers to animal
Pereoglou, F., Population processes in an early successional heathland 71 Chapter 4: Landscape genetics movements (Scribner et al. 2005). Current research on landscape connectivity has focussed on the influence of ‘habitat breaks’ (i.e. anthropogenic landscape fragmentation) (Gauffre et al. 2008; Macqueen et al. 2008; Spear & Storfer 2008; Pavlacky et al. 2009), natural barriers (e.g. rivers, mountains and topography) (Spear et al. 2005; Murphy et al. 2010), or a combination of both (Cushman et al. 2006; Perez-Espona et al. 2008; Short Bull et al. 2011). There is limited information on the effect of natural disturbance such as wildfire, naturally heterogeneous landscapes, or natural variation in habitat suitability on patterns of genetic diversity, population structure and gene flow. Fire is a major form of natural disturbance (Turner et al. 1998). Different fire regimes produce distinct landscape patterns and species arrays (Bond & Keeley 2005). Population recovery and persistence in post-fire environments is influenced by fire regimes, and their effects on landscape features including patch isolation and connectivity (Clarke 2008). With altered fire regimes and increased fire occurrence being a likely impact of climate change (Williams et al. 2001; Flannigan et al. 2009; Steffen et al. 2009), our understanding of how fire events influence biota is a critical contemporary environmental issue. Fire playing a positive role in maintaining habitat suitability is an effect that is plausible for post-fire specialist fauna and those that colonise early successional ecosystems, but one that is not well studied. In this paper, we examined the population genetic structure of the eastern chestnut mouse (Pseudomys gracilicaudatus Gould 1845), a threatened small mammal endemic to the east coast of Australia and dependent on regular fire. We addressed a series of hypotheses relating to the dispersal strategy of a post-disturbance specialist mammal, including: (1) Fire/burned areas will promote gene flow. Several studies have shown that fire affects genetic structure through environmental changes that cause demographic variation such as population expansion (including colonisation) or decline, and variation in connectivity (Templeton et al. 2001; Schrey et al. 2010, 2011; Templeton et al. 2011). For species that specialise in early seral stage, post-disturbance habitats, we expect a reduction in genetic structure in recently burned areas. This is because fire increases the area of suitable habitat, so small and relatively isolated pre-fire populations will expand and individuals will disperse to colonise newly-suitable habitat. Population expansion, increased connectivity between existing populations and colonisation of new areas (particularly with a large number of colonists) are expected to weaken genetic structure (Wade & McCauley 1988; Templeton et al. 2001; Excoffier et al. 2009). Our prediction may be incorrect if the number of individuals colonising newly suitable habitat after fire is very small and ongoing migration in subsequent generations is rare (Wade & McCauley 1988). However, our hypothesis was that the former prediction would hold. (2) Natural habitat heterogeneity will not restrict gene flow. Habitat suitability has been considered as a correlate of ‘conductance’ to dispersal across a landscape, on the assumption that environmental types that rank poorly by occupancy or abundance criteria will also be poor Pereoglou, F., Population processes in an early successional heathland 72 Chapter 4: Landscape genetics for dispersal (Spear et al. 2010, e.g. Wang et al. 2008). The eastern chestnut mouse has a naturally patchy distribution within our study area due to very specific requirements for vegetation type and fire history (Pereoglou et al. 2011). Habitat suitability is thus spatially patchy and temporally dynamic. We predicted that an effective dispersal strategy to access and colonise suitable habitat under such dynamic environmental conditions would be one in which dispersal is possible across all landscape types and is not restricted by perceived habitat preferences. (3) Dispersal will not be strongly sex-biased. Male-biased dispersal is common among mammals (Greenwood 1980) and occurs in other small mammal species that occur in our study area, including the brown antechinus (Antechinus stuartii) and the bush rat (Rattus fuscipes) (Cockburn et al. 1985; Peakall et al. 2003). However, we predicted a lack of strong sex-bias in dispersal for an early seral specialist. Modelling suggests that strong sex-biased dispersal is not favoured under high demographic stochasticity (Gros et al. 2008). After fire, dispersal by eastern chestnut mice is likely to be associated with colonisation of newly suitable habitat, and greater colonisation ability by males compared to females would be disadvantageous in that it would increase the likelihood of settling in unoccupied habitat with no potential mates.
4.3 Methods
4.3.1 Study species
Pseudomyine rodents provide some of the best examples of fire specialist fauna (Catling & Newsome 1981) due to their rapid establishment and early breeding after fire, rapid population increase, and ability to maintain viable breeding populations from first colonisation (Fox 1982). The eastern chestnut mouse is an early seral specialist that colonises heathland habitat, peaking in abundance post-fire before declining over time as vegetation regenerates (Fox 1982; Monamy & Fox 2000; Fox et al. 2003). It has a disjunct distribution along the east coast of Australia and occurs at low density throughout much of its range. The species was considered extinct in southern Australia until 2002 when it was discovered in the Jervis Bay region (Meek & Triggs 1997). This area now supports the only contemporary population of the eastern chestnut mouse south of Gosford, New South Wales (33°23'S 151°18'E) and is arguably the most isolated population of the species in eastern Australia. The eastern chestnut mouse is an annual summer breeder capable of producing multiple litters of one to five (usually three) offspring (Fox 2008). The species lives for less than two years in wild populations (Watts & Kemper 1989). Early development is rapid which contrasts with a slow growth rate to reach maturity (Fox & Kemper 1982). Adults show significant sexual dimorphism (Fox & Kemper 1982) and attain average weights of 90 g for males and 75 g for females (Fox 2008). Little is known about social organisation and behaviour in this species. Radio-tracking data suggests that adult eastern chestnut mice are solitary, territorial and Pereoglou, F., Population processes in an early successional heathland 73 Chapter 4: Landscape genetics maintain female home ranges of 0.5–1 ha and up to 4 ha for males (Pereoglou et al. unpublished data).
4.3.2 Study area and sampling
We collected tissue from 26 sites surveyed between 2008 and 2010 in Booderee National Park, located approximately 200 km south of Sydney, Australia (Fig. 4.1). We genotyped 287 individual animals and sample sizes ranged from 1 to 58 mice per site (Fig. 4.1). We used two approaches to survey the study area. Initial site selection was randomly stratified by vegetation type and consisted of a single 100 m transect. Each transect comprised: (i) a black plastic drift fence connecting 380 mm deep x 300 mm diameter pitfall traps at 0–20 m, 40–60 m, and 80–100 m; (ii) large wire cage traps (30 x 30 x 60 cm) placed at 0 m and 100 m with four small wire cage traps (20 x 20 x 50 cm) at 20 m intervals between; and (iii) ten aluminium box traps (Elliot Scientific Equipment, Australia) placed every 10 m along the transect line from 0–90 m. Data on the number of transects in each vegetation type, the number with eastern chestnut mouse captured, and the corresponding occupancy estimates are given in Table S1 (in Appendix 5). We then chose five transects and selected an additional four sites for intensive grid trapping to sample sufficiently high densities of the eastern chestnut mouse to allow for population genetic analyses. We classified all grid sites as dry heath (Taws 1997) and each had been burned by wildfire in December 2003. Our grid arrangement comprised lines at 20 m intervals with traps 10 m apart covering an area ranging from 1–3 ha (1.9 0.19 ha). Transect sites were surveyed annually during the eastern chestnut mouse breeding season (November – February) and grid sites were surveyed twice each year, during the breeding season and also during the non-breeding season (June – August). We opened traps for three or four consecutive nights (31 548 trap nights in total) and baited with rolled oats, peanut butter and fennel seed. We assessed captured animals for weight, sex, age cohort, breeding condition, and collected a 2 mm ear biopsy stored in 100% ethanol.
4.3.3 DNA extraction, PCR amplification and genotyping
We extracted DNA from ear tissue using the method of (Bruford et al. 1988) and re- suspended the DNA pellet in 100µl of TE buffer (10 mM Tris and 0.1 mM EDTA). We genotyped samples at 10 microsatellite loci (Moro & Spencer 2003) (Table S2). Microsatellite PCRs with a total volume of 30 µL contained 10X buffer (Qiagen), 200 µM dNTPs, 1.8 mM MgCl2, 0.5 µL BSA, 4 pmoles reverse primer, 1.5 pmoles forward tailed primer, 2.5 pmoles fluorescent end-labelled M13 primer sequence (Schuelke 2000), 0.2 units Taq polymerase (Qiagen), and 1.5 µL template DNA. We ran all reactions on a thermal cycler (Applied Biosystems, Inc.) with initial denaturation of 94–95o C for 3 min, then cycled using PCR conditions described in Table S2, followed by a final extension step of 72o C for 10 min. We ran the PCR products on an ABI 3130 (Applied Biosystems Inc.) automated sequencer with the LIZ
Pereoglou, F., Population processes in an early successional heathland 74 Chapter 4: Landscape genetics 500 size standard. We ran negative controls in PCRs and in each sequencer run to check for contamination. We used GeneMapper Version 3.7 (Applied Biosystems Inc.) to score each locus for each sample.
Figure 4.1: Study area and distribution of survey sites. Solid symbol indicates site with eastern chestnut mouse presence and the number of genotyped individuals is provided above symbol. Hollow symbol indicate sites with no eastern chestnut mouse captures
4.3.4 Genetic data analysis
We used the genotype matching function of MSTOOLS (Park 2001) to identify identical genotypes. Duplicate genotypes were re-labelled if the corresponding capture records were consistent with multiple captures and samples from the same individual (e.g. due to temporary marking or tag loss). We ran analyses separately for each survey year. Survey years included a twelve-month period from the beginning of the eastern chestnut mouse breeding season in September to August of the following year (September–August 2008-09 and September–August 2009-10).
4.3.4.1 Population genetic summary statistics We excluded locations from the population-level analyses if sites had less than 10 individuals genotyped. We calculated standard population genetic summary statistics for all loci and sites including the number of alleles, observed and expected heterozygosities, and FIS using
Pereoglou, F., Population processes in an early successional heathland 75 Chapter 4: Landscape genetics GENALEX version 6.3 (Peakall & Smouse 2006). We tested for departures from Hardy- Weinberg equilibrium using FSTAT (Goudet 2001).
4.3.4.2 Population subdivision and isolation by distance (IBD)
We used AMOVA to estimate population subdivision and calculated pairwise values of FST for all sites where sample sizes were greater than 10 individuals. We used Mantel tests to test for isolation by distance on pairwise FST between sites (FST /1- FST vs log geographic distance). Unless otherwise stated, all permutational tests were performed 999 times and analyses were completed using GENALEX. To further investigate the spatial pattern of genetic structure, we calculated the spatial autocorrelation (r) of pairwise multilocus genetic distances over a set of geographic distance classes (Smouse & Peakall 1999). We calculated pairwise multilocus genetic distances (GD) according to the methods of Smouse & Peakall (1999) and tested the null hypothesis of no spatial genetic pattern (r = 0) against the alternative that spatial autocorrelation was positive at some distances and negative at others (r ≠ 0). For each survey year (2008-09 n = 172, 2009-10 n = 141), we ran two different analyses. In the first analysis we selected 500 m distance classes up to 5 km. The second was a finer-scaled analysis, where we investigated spatial autocorrelation at 50 m intervals up to 250 m (i.e. fine scale structure within sampling sites). The choice of distance classes primarily relates to our sampling scheme, where the first (coarse-scaled) distance classes were larger than our sampling sites so we were testing differences between spatial autocorrelation among individuals within sites (0-500m) and at increasing distances between sites. The finer-scaled analysis focussed on testing for spatial clustering of similar genotypes within the scale of our sampling grids. We used 999 permutations with a bootstrap re-sampling procedure run 1000 times to estimate the 95% confidence interval. We used the programs STRUCTURE version 2.3 (Pritchard et al. 2000) and GENELAND version 3.2.4 (Guillot et al. 2005) to complement our other analyses of population structure with the specific aim of testing whether our samples could be clustered into spatially discrete sub- populations (parameters in Appendix S1 in Appendix 5). We used the program GESTE v2 (Foll & Gaggiotti 2006) to estimate effects of landscape variables on population-specific FST. For this analysis, we restricted dataset to nine sites with the greatest sample sizes, from which a mean of 28 individuals was sampled (s.d. = 16). All of these sites were within areas burnt in 2003, so we were unable to test for effects of fire history. However, we tested for effects of vegetation heterogeneity with variables representing the proportion of heath (the most-preferred vegetation type) within 200m and 500m buffers, as well as sample size (as a proxy for local population density), and X and Y coordinates to test for broad spatial gradients in the strength of genetic structure (e.g. an east-west peninsula effect). For this analysis, we pooled individuals from both years within the same sites. Many individuals
Pereoglou, F., Population processes in an early successional heathland 76 Chapter 4: Landscape genetics were sampled in both years and previous analyses showed no major differences in the patterns detected in either year.
4.3.4.3 Sex-biased dispersal To test the hypothesis that genetic structure and corresponding philopatry is not sex-biased we used two approaches. First, we used the heterogeneity test of Smouse et al. (2008) to compare the spatial autocorrelation r-values of males and females with the multi-populations procedure in GENALEX. We assessed spatial autocorrelation at two scales: 500 m distance classes up to 5 km and 50 m distance classes up to 250 m. If there was a sex-bias, for the philopatric sex we expected declining correlations of r-values with increasing distance if the null hypothesis of r = 0 was not correct. If both sexes showed a similar pattern of decline, we expected the r-values in the first distance would be largest for the philopatric sex (Smouse et al. 2008). We focussed on the first distance class because if one sex remains relatively philopatric (such that relatives are spatially clustered), any differences in spatial autocorrelation between the sexes are expected to be most apparent in this class (Banks & Peakall 2012). We used 999 permutations with a bootstrap re-sampling procedure run 1000 times to estimate the 95% confidence interval. Sample size differed by survey year: 2008-09 male n = 79, female n = 90, and 2009-10 male n = 65, female n = 76. Secondly, we used a set of one-sided biased dispersal tests in FSTAT (Goudet et al. 2002). We permuted these tests 10000 times to compare the estimated metrics including mean corrected assignment index, variance of corrected assignment index, FST, FIS and HS between male and female animals. Sample size differed by survey year: 2008-09 n = 160 animals in 12 sites and 2009-10 n = 132 animals in 10 sites. Sites without adequate sample sizes of both sexes were excluded from analysis in FSTAT.
4.3.4.4 Landscape resistance We used partial Mantel tests in a causal modelling framework to analyse the influence of habitat on gene flow and to determine the extent to which possible landscape resistance models explained the spatial pattern of genetic distances between individuals. We created a matrix of genetic distances based on the same genetic distance measure used for the spatial autocorrelation analysis described above (Smouse & Peakall 1999). We produced landscape resistance surfaces representing different hypotheses about the difficulty in traversing different habitat types using CIRCUITSCAPE version 3.5 (McRae & Shah 2009). CIRCUITSCAPE uses circuit theory (McRae & Beier 2007; McRae et al. 2008) to predict connectivity in heterogeneous landscapes and allows for multiple pathways of gene flow rather than a single corridor with user-defined dimensions. We used the pairwise source/ground modelling mode to calculate connectivity between genotype point locations with habitat data specifying per cell conductance values, an eight neighbours cell connection, and calculations based on average conductance.
Pereoglou, F., Population processes in an early successional heathland 77 Chapter 4: Landscape genetics We assigned conductance values to raster cells in ARCGIS version 9.2 (ESRI 2009) representing hypotheses about the influence of distance, fire and vegetation type on genetic connectivity (Fig. 4.2; Table 4.1). (1) Distance: Our first hypothesis and simplest model was a test of isolation by distance across a uniform-resistance terrestrial landscape (Wright 1943). We did not use the classic Euclidean distance measure as it was inappropriate to represent straight line distances given the study area was a peninsula and the coastline and ocean create a substantial barrier for a terrestrial small mammal. Therefore, in our null model we assumed movement could occur in any direction, all raster cell values were equal in conductance, and the only limit to gene flow was that imposed by the coastline and ocean. (2) Fire: Our second hypothesis was that fire promotes genetic connectivity for fauna that specialise in early successional ecosystems. We expected to find a positive effect of fire (i.e. a relative negative effect of unburnt areas) because it resets the vegetation succession creating suitable habitat for post-fire specialist species (Monamy & Fox 2000). Residual survivors become responsible for post-perturbation population recovery (Turner et al. 1998; Banks et al. 2011) potentially following burned vegetation to recolonise and reproduce. Because causal modelling provides only significance tests and does not yield estimates of resistance from which we could assess the effect of fire, we took the approach of testing a series of models covering what we considered to be the range of plausible conductance values of fire (Table 4.1). (3) Vegetation type: Our third hypothesis was that for species with spatially and temporally variable habitat requirements, movement across unsuitable vegetation types would need to occur to access newly created habitat patches post-fire. In our study, the eastern chestnut mouse was highly habitat specific in occurrence across a landscape characterised by a heterogeneous mosaic of vegetation types. Therefore, we took two approaches to test the influence of vegetation type on genetic connectivity: firstly we classified vegetation types as habitat versus non-habitat based on whether they were occupied or not and parameterised the models according to the range of plausible conductivity values (with the conductance value of habitat ranging from equal to non- habitat up to 1000 times greater than non-habitat); and secondly we considered conductivity was proportional to habitat quality as assessed by the percentage of sites of each vegetation type that were occupied (Table 4.1). We proposed nine alternative landscape models (Table 4.1) that we tested against the ‘null landscape’ isolation by distance hypothesis. We used Mantel and partial Mantel correlations as the tests statistics for causal modelling, including (1) simple Mantel tests between genetic distance and landscape resistances; (2) partial Mantel tests between genetic distance and landscape resistances, partialling out the effects of the null model; and (3) for those landscape resistance scenarios for which (2) was significant, partial Mantel tests between genetic distance and the null model, partialling out the effects of the landscape resistance models. We conducted Mantel and partial Mantel tests using the ECODIST package (Goslee & Urban 2007) in R version 2.7 (R Development Core Team 2008) with 10000 randomisations. To infer an effect of
Pereoglou, F., Population processes in an early successional heathland 78 Chapter 4: Landscape genetics a landscape resistance scenario on dispersal, we expected (1) and (2) to be significant, and we expected (3) to be non-significant if that scenario ‘correctly’ explained population connectivity in our study population (Cushman et al. 2006). As an alternative statistical test of the effect of landscape resistance after accounting for the ‘null’ model, we used multiple regression on distance matrices (Lichstein 2007).
Table 4.1: Raster cell values used to create resistance landscapes. Water was assigned a value of zero
Model Hypothesis Cell conductance values Null All habitat types exert the same level of All cells = 1 permeability to gene flow Firemodel1 Areas burnt in the 2003/4 fire are twice Burnt areas = 2 as permeable to gene flow as unburned All other cells = 1 habitat Firemodel2a Areas burnt in the 2003/4 fire are 10 Burnt areas = 10 times as permeable to gene flow as All other cells = 1 unburned habitat Firemodel2b Areas burnt in the 2003/4 fire are 100 Burnt areas = 100 times as permeable to gene flow as All other cells = 1 unburned habitat Firemodel2c Areas burnt in the 2003/4 fire are 1000 Burnt areas = 1000 times as permeable to gene flow as All other cells = 1 unburned habitat Vegmodel3† Vegetation types where mice have been Cells in heathland, woodland, and sedgeland captured are twice as likely to facilitate = 2 gene flow Cells in other vegetation types =1 Vegmodel4a† Vegetation types where mice have been Cells in heathland, woodland, and sedgeland captured are 10 times as likely to = 10 facilitate gene flow Cells in other vegetation types =1 Vegmodel4b† Vegetation types where mice have been Cells in heathland, woodland, and sedgeland captured are 100 times as likely to = 100 facilitate gene flow Cells in other vegetation types =1 Vegmodel4c† Vegetation types where mice have been Cells in heathland, woodland, and sedgeland captured are 1000 times as likely to = 1000 facilitate gene flow Cells in other vegetation types =1 Vegmodel6† Permeability to gene flow is associated Cells in heathland, woodland, and sedgeland with habitat quality = percentage of surveyed sites of that vegetation type that were occupied (1-100) Cells in other vegetation types =1 † Indicates cell values were estimated from unpublished annual monitoring data (refer to Table S1 in Appendix 5)
Pereoglou, F., Population processes in an early successional heathland 79 Chapter 4: Landscape genetics
Figure 4.2: Habitat quality relating to movement across a null landscape (A), burned areas (B), selected vegetation types (C), and occupied habitat (D). Gray-scale based on cell conductance values listed in Table 1. Darker colour indicates a lower resistance value assigned
Pereoglou, F., Population processes in an early successional heathland 80 Chapter 4: Landscape genetics 4.4 Results
4.4.1 Population genetic summary statistics
Genotype proportions deviated from H-W expectations in 2008-09 at locus 7D12 in site 7-P, and in 2009-10 at locus 9A8 in site 7-I, locus 10G6 in site 7-46, and locus 1A7 in site 7-95. Locus 6D4 showed consistent significant departures from HWE (in sites 7-44B, 7-2 and 7-H in 2008-09 and at 7-44B, 7-46 and 7-M in 2009-10) and was removed from subsequent analyses. With the exception of locus 6D4, the departures from H-W expectations in our dataset were not beyond the false error rate for multiple testing (i.e. seven out of 130 tests were likely to be significant), nor were they consistent across sites or years. Overall, we identified between three and six alleles per locus. Genetic diversity across sites ranged from 0.499–0.652 in 2008-09 and 0.415–0.632 in 2009-10, with observed heterozygosity averaged across sites greater than expected in both survey years (Table 4.2).
4.4.2 Population subdivision and spatial genetic structure
We found significant genetic differentiation among sites across the study area (2008-09
FST=0.161, p=0.001; 2009-10 FST =0.195, p=0.001) (Table S3 in Appendix 5) and an effect of isolation by distance. Euclidean distance (log-scale geographic distance) was a significant predictor of FST between sites (2008-09 Mantel r=0.703, p=0.007; 2009-10 Mantel r=0.377, p=0.044 r=0.377, p=0.044) (Fig. 4.3). At a broad scale, we detected positive and significant spatial autocorrelation in distance classes up to 2 km (Fig. 4.4A, B). At distances exceeding 3–3.5 km, pairwise multilocus genotypes were less similar than those assigned to distance classes at random. At a fine scale, the within-site distribution of genotypes was significant and positively autocorrelated up to 200–250 m (Fig. 4.4C, D). We did not identify discrete, spatially segregated, genetic populations above the scale of sampling grids using the clustering methods. STRUCTURE did identify the presence of three groups in both years. However, it could not assign individuals to a population with > 90% confidence (Table S5 in Appendix 5). Further, these groups did not correspond to spatially discrete clusters of individuals and we found that samples from a particular site included a mixture of individuals from different STRUCTURE groups (Appendix S2 and Figure S1 in Appendix 5). GENELAND identified seven groups in 2008-09 and eight in 2009-10. Individuals were assigned to clusters that reflected population subdivision based on pairwise
FST, suggesting field sites comprised distinct demographic units (Appendix S2 and Table S4 in Appendix 5). However, GENELAND could not assign individuals to a population with > 90% confidence (Table S5 in Appendix 5). Broadly, GENELAND results matched with pairwise FST between sites, but did not add any higher-level spatial groupings of individuals.
Pereoglou, F., Population processes in an early successional heathland 81 Chapter 4: Landscape genetics
Table 4.2: Population genetic parameters for the eastern chestnut mouse. The number of samples (n), the number of alleles (Na), observed heterozygosity (HO), expected heterozygosity (HE) and FIS for each site averaged across loci ± standard error
2008-09 2009-10 Site n Na HO HE FIS N Na HO HE FIS 7-44 8 2.9 ± 0.2 0.597 ± 0.093 0.589 ± 0.028 0.005 ± 0.134 0 7-46 38 3.8 ± 0.3 0.610 ± 0.069 0.567 ± 0.058 -0.070 ± 0.037 24 3.2 ± 0.2 0.585 ± 0.048 0.551 ± 0.042 -0.075 ± 0.055 7-95 2 11 3.0 ± 0.2 0.524 ± 0.068 0.500 ± 0.044 -0.029 ± 0.073 7-2 23 3.1 ± 0.3 0.528 ± 0.078 0.471 ± 0.065 -0.112 ± 0.037 18 3.1 ± 0.2 0.530 ± 0.069 0.526 ± 0.052 -0.002 ± 0.069 7-44B 21 3.3 ± 0.2 0.579 ± 0.038 0.556 ± 0.039 -0.048 ± 0.031 17 3.3 ± 0.2 0.572 ± 0.024 0.569 ± 0.028 -0.021 ± 0.055 7-H 23 2.9 ± 0.3 0.499 ± 0.058 0.462 ± 0.055 -0.093 ± 0.040 21 2.9 ± 0.2 0.415 ± 0.055 0.434 ± 0.055 0.032 ± 0.055 7-I 19 3.3 ± 0.2 0.567 ± 0.045 0.569 ± 0.020 0.008 ± 0.066 14 3.1 ± 0.2 0.492 ± 0.067 0.473 ± 0.060 -0.018 ± 0.093 7-M 3 13 2.9 ± 0.2 0.632 ± 0.028 0.519 ± 0.038 -0.246 ± 0.059 7-P 12 3.0 ± 0.2 0.652 ± 0.067 0.555 ± 0.039 -0.172 ± 0.093 8 2.8 ± 0.2 0.621 ± 0.062 0.520 ± 0.048 -0.203 ± 0.077 Mean 3.2 ± 0.1 0.576 ± 0.025 0.538 ± 0.018 -0.069 ± 0.027 3.0 ± 0.1 0.547 ± 0.020 0.511 ± 0.016 -0.070 ± 0.025
Pereoglou, F., Population processes in an early successional heathland 82 Chapter 4: Landscape genetics
Figure 4.3: Relationship between genetic distance (pairwise site FST) and geographic distance. Solid symbol is 2008-09 data; hollow symbol is 2009-10 data
Analyses in GESTE v2 identified a mean population-specific FST estimate of 0.105 (s.d. of 0.044 across sites) (Figure S2 in Appendix 5). The constant-only model was best supported (posterior probability = 0.727). This suggests that none of the modelled variables representing vegetation heterogeneity or spatial location affected pop-specific FST.
4.4.3 Sex-biased dispersal
We found some evidence that spatially proximal females were more genetically similar than males in one of our analyses in on year of sampling, however this pattern was inconsistent. We did not find strong evidence for heterogeneity between correlograms for males and females. In 2008-09, female r-values were significantly greater than males in the 0–50 m distance class according to the T2 test (T2=4.182, p=0.033) but not in the 0–500 m distance class (T2=0.001, p=0.977). However, the bootstrap 95% CIs of the r values, a more conservative test (Banks & Peakall 2012) for males and females overlapped (Table S6 in Appendix 5). In 2009-10, female r-values were not significantly greater than males in the 0–50 m distance class (T2=0.087, p=0.751) but they were in the 0–500 m distance class (T2=14.258, p =0.001). This was the only spatial autocorrelation value test of sex-biased dispersal to be supported by the conservative bootstrap 95% CI overlap test (Table S6 in Appendix 5). There were no consistent biological patterns in the other distance class comparisons in either survey year (Table S6 in Appendix 5).
Pereoglou, F., Population processes in an early successional heathland 83 Chapter 4: Landscape genetics
Figure 4.4: Correlograms indicating spatial autocorrelation of pairwise multilocus genetic distances over 5 km (A) 2008-09, (B) 2009-10; and over 250m (C) 2008-09, (D) 2009-10. Broken lines represent the 95% CI for the null hypothesis of a random distribution of genotypes. Error bars represent the 95% CI around r for a specific distance class.
None of the sex-biased dispersal tests in FSTAT provided evidence for female philopatry and male dispersal and none of the test statistics were significant in either survey year (Table S7 in Appendix 5). The test statistics were conflicting in regards to expectations of male-biased dispersal (e.g. FST, FIS and HS; Table S7 in Appendix 5). The lack of apparent sex-biased
Pereoglou, F., Population processes in an early successional heathland 84 Chapter 4: Landscape genetics dispersal may be attributed to small sample sizes used in these analyses. However, given that sex-biased dispersal can be detected with similar sample sizes in species where this is a prevailing life history strategy, such as Antechinus species (Banks & Peakall 2012), our results are likely due to biological reasons discussed in the next section.
4.4.4 Landscape resistance (1) Simple Mantel tests between genetic distance and landscape resistances. All of the simple Mantel tests were significant (Table 4.3). Both the null model and the alternate resistance landscapes explained variation in genetic distance between individuals. However, when the models were ranked based on Mantel r, only Firemodel1 performed better than the null model. (2) Partial Mantel tests between genetic distance and landscape resistances, partialling out the effects of the null model. We found significant effects of the fire models 1, 2a, 2b and 2c (conductance across burned habitat was twice, 10, 100 and 1000 times greater than unburned habitat, respectively; Table 4.1) after accounting for variation in genetic distance explained by the null landscape. After we accounted for the null model, the partial correlations of the vegetation models (with increases in relative conductivity of preferred habitat to other vegetation types and with conductance parameterised to reflect the distribution of capture records and habitat preferences) with genetic distance were non-significant (Table 4.3).
(3) Partial Mantel tests between genetic distance and the null model, partialling out the effects of the landscape resistance models. Of the landscape resistance scenarios that explained a significant component of the variation in genetic distance between individuals when the null model was partialled out (the four fire models: Table 4.1), the null model was non-significant after Firemodel1 was accounted for (Table 4.3). This model featured a conductance of recently burned habitat twice that of unburned habitat. The null model explained a significant component of the variation in genetic distance even after the remaining fire models (Firemodel2a-c, featuring relatively higher conductance values of burned habitat: Table 4.1) were fitted (Table 4.3). This suggests that the latter models potentially overestimate the relative conductivity of burnt habitat.
4.5 Discussion
Our study provides support for a ‘generalist’ dispersal strategy, which would be advantageous for a species that prefers ephemeral habitat, conditional on spatial and temporal variability. We found significant genetic differentiation between sites that followed an isolation by distance pattern, and significant spatial autocorrelation between genotypes, indicating dispersal was restricted relative to the overall scale of the study. However, we did not find strong evidence for dispersal to be male-biased and limited to only one sex. Our data suggested no effect of heterogeneity in heterogeneous native vegetation on genetic structure, despite
Pereoglou, F., Population processes in an early successional heathland 85 Chapter 4: Landscape genetics strong habitat selection for specific vegetation types, and there was evidence for a positive effect of fire in promoting connectivity.
Table 4.3: Results of Mantel and partial Mantel tests of landscape resistance on gene flow. GD = pairwise multilocus genetic distance. Data for both survey years is combined. Significance levels from multiple regression on distance matrices (MRDM) are included for comparison
Mantel test MRDMMRDM Matrix 1 Matrix 2 Partial rM p-value p-value Matrix 1 Matrix 2 Partial rM p-value p-value GD NullNull 0.2150.215 <0.001<0.001 <0.001<0.001 GD firemodel1firemodel1 0.2290.229 <0.001<0.001 <0.001<0.001 GD firemodel2a 0.1960.196 <0.001<0.001 <0.001<0.001 GD firemodel2b 0.1220.122 0.0090.009 0.0060.006 GD firemodel2c 0.1100.110 0.0120.012 0.0120.012 GD vegmodel3 0.2150.215 <0.001<0.001 <0.001<0.001 GD vegmodel4a 0.1940.194 <0.001<0.001 <0.001<0.001 GD vegmodel4b 0.1370.137 <0.001<0.001 <0.001<0.001 GD vegmodel4cegmodel4c 0.1170.117 <0.001<0.001 <0.001<0.001 GD vegmodel6 0.1570.157 <0.001<0.001 <0.001<0.001 (2) Partial Mantel tests between genetic distance and landscape resistances, partialling out the effects of the null model GD firemodel1firemodel1 nullnull 0.0850.085 0.0280.028 0.0260.026 GD firemodel2a nullnull 0.0840.084 0.0320.032 0.0280.028 GD firemodel2b nullnull 0.0840.084 0.0330.033 0.0340.034 GD firemodel2c nullnull 0.0160.016 0.5240.524 0.5110.511 GD vegmodel3 nullnull 0.0380.038 0.1800.180 0.1740.174 GD vegmodel4a nullnull -0.050-0.050 0.0610.061 0.0680.068 GD vegmodel4b nullnull -0.054-0.054 0.0530.053 0.0550.055 GD vegmodel4c nullnull -0.042-0.042 0.1120.112 0.1090.109 GD vegmodel6 nullnull -0.033-0.033 0.3220.322 0.3160.316 (3) Partial Mantel tests between genetic distance and the null model, partialling out the effects of the null landscape resistance models GD firemodel1firemodel1 nullnull -0.029-0.029 0.4050.405 0.3080.308 GD firemodel2a nullnull 0.1250.125 <0.001<0.001 <0.001<0.001 GD firemodel2bl2b nullnull 0.1970.197 <0.001<0.001 <0.001<0.001 GD firemodel2c nullnull 0.2040.204 <0.001<0.001 <0.001<0.001 GD vegmodel3 nullnull 0.0110.011 0.6140.614 0.6050.605 GD vegmodel4a nullnull 0.0980.098 <0.001<0.001 <0.001<0.001 GD vegmodel4b nullnull 0.1750.175 <0.001<0.001 <0.001<0.001 GD vegmodel4c nullnull 0.190.19 <0.001<0.001 <0.001<0.001 GD vegmodel6 nullnull 0.1540.154 <0.00<0.001 1 <0.001<0.001
4.5.1 Landscape connectivity and gene flow For a pyric specialist species with preference for early successional stage vegetation, we predicted fire would promote gene flow and enhance connectivity because following burned vegetation would potentially lead dispersers to unoccupied suitable habitat. While we identified significant genetic differentiation between populations, we found no evidence for discrete structure above the level of sampling sites (i.e. broad spatial population groupings, or clusters), suggesting that a model of isolation by distance is an appropriate interpretation of this system.
The population-specific FST analysis in GESTE provided no evidence for genetic structure being influenced by the distribution of preferred and non-preferred habitat (vegetation heterogeneity).
Pereoglou, F., Population processes in an early successional heathland 86 Chapter 4: Landscape genetics Consistent with this finding, landscape genetics analyses of the relationships between pairwise individual genetic distances and a set of landscape resistance scenarios did not support an effect of vegetation-mediated habitat suitability on spatial genotypic patterns. This suggests that occupancy and abundance are not appropriate surrogates for landscape permeability to movement in disturbance adapted species (see Spear et al. 2010). However, these analyses did support an increased relative permeability of burnt landscapes over unburnt landscapes, suggesting that fire promotes connectivity in this species. The support for increased permeability of burnt relative to unburnt habitat came from simple Mantel tests of inter-individual genetic versus landscape resistance distances, in which Firemodel 1 (Table 4.1), where the conductance of recently-burned habitat was twice that of unburned habitat, had the strongest simple Mantel correlation (Table 4.3). Although our Mantel correlations were not high (maximum r = 0.229), we expected that geographical correlations with individual multilocus distances to be ‘messier’ than population-based distances as the latter do not include intrapopulation (inter-individual) variation. Further, partial Mantel tests and multiple regression on distance matrices (MRDM) identified this model as being significantly associated with genetic distance after accounting for a ‘null’ landscape model. This model that satisfied the classic causal modelling criteria for acceptance (see Cushman et al. 2006) was Firemodel1 (Table 4.1), in which the conductance of recently-burned habitat was twice that of unburned habitat. Landscape resistance scenarios in which recently burned habitat had a conductance 10, 100 or 1000-fold greater than unburned habitat were significant after accounting for the null model, but the null model was also significant after these models were accounted for. This suggests that these latter models overstate the relative conductance value of recently-burned habitat and that Firemodel1 is a reasonably accurate representation of the resistance landscape for the studied population. While this finding was consistent with our prediction for this species, and with other lines of evidence (the observed heterozygosity excess in both years is consistent with increased dispersal and admixture of populations that were likely to have been considerably smaller and more isolated before the 2003 fire), we do interpret the partial Mantel and MRDM analyses with some caution, as recent simulation work has shown that these tests can, under some circumstances, have unacceptable error rates for the significance of spatially-structured explanatory variables (Jaquiéry et al. 2011, Guillot & Rousset 2012). Statistical caveats aside, we discuss the implications of fire for connectivity in post-fire specialists below. Potentially, fire promotes connectivity for early successional specialist fauna in two ways. First, fire improves habitat suitability for such species and habitat suitability improves landscape conductance to gene flow. Where long-term fire suppression occurred in a previously fire- maintained ecosystem, Templeton et al (2001) observed a significant increase in dispersal and colonisation rate of new patches by collared lizards after re-initiation of burning. The increase in dispersal was attributed not to the fire itself but to the change in structure of the intervening
Pereoglou, F., Population processes in an early successional heathland 87 Chapter 4: Landscape genetics habitat matrix and opening of the understorey (Templeton et al. 2001). For example, burned areas may not be an impediment to movement as long as some structure, native plant regeneration, or patches of remnant vegetation are present. A post-fire collapsed shrub layer may even provide cover and protection facilitating movement (Spear et al. 2005). Spear and Storfer (2010) found gene flow was maintained across burned areas and suggested movement was facilitated by biological legacies (sensu Franklin et al. 2000) in the form of downed trees and debris. For dispersing individuals of the eastern chestnut mouse, a fire boundary of unburned habitat or late successional stage vegetation might be a greater deterrent to movement than the burnt landscape (i.e. habitat cues can assist directional movement; Bowler & Benton 2005). Under this scenario, the underlying mechanism for persistence of post-fire specialist fauna may be the disperser’s ability to follow fire in an attempt locate suitable habitat. Second, an alternative explanation is that the resistance landscape for dispersal by the eastern chestnut mouse is not static, and that dispersal is increased dramatically in burnt areas in association with colonisation immediately after the fire, then returns to levels described under a 'null landscape' model. Genetic signals of recent (but not current) dispersal scenarios can persist for a number of generations (Landguth et al. 2012). Under this scenario, fire causes substantial mortality but also results in rapid colonisation of newly created early successional habitat by residual survivors and thus increased gene flow over the immediate short-term post-fire. Once the empty territories have been colonised, there is no longer any reason for higher dispersal across burnt areas. Our data do not distinguish these two explanations. For species with spatially and temporally variable habitat requirements, we predicted natural habitat heterogeneity would not restrict gene flow because dispersal would still need to occur through landscapes which do not meet specific habitat requirements. A heterogeneous cover of native vegetation types is unlikely to restrict gene flow because dispersal is still possible through landscapes which are considered inappropriate when conditional on perceived habitat preferences. Although the eastern chestnut mouse has specific habitat requirements including early successional stage heathland vegetation, we found genetic connectivity across the study area was not limited by vegetation type and that other native vegetation communities interspersed throughout heathland did not restrict movement. While barriers to dispersal may funnel movement in one direction (Gustafson & Gardner 1996), dispersal through unfavourable habitat has been recorded for many species (Ruscoe et al. 1998; Mossman & Waser 2001; Selonen & Hanski 2004; Gauffre et al. 2008; Macqueen et al. 2008). Selectivity for preferred habitat during dispersal has also been recorded for other species (Lorenz & Barrett 1990; Baur & Baur 1992; Schultz 1998; Lindenmayer & Peakall 2000; Banks et al. 2005), indicating the effect (or not) of habitat heterogeneity on gene flow is likely to be strongly species-specific and/or site-specific. A species’ capacity for movement through heterogeneous landscapes is likely to depend on whether heterogeneity is provided by indigenous vegetation of varying composition (as demonstrated in our study) or by small, remnant patches of native vegetation
Pereoglou, F., Population processes in an early successional heathland 88 Chapter 4: Landscape genetics interspersed with anthropogenic modified landscapes (i.e. agricultural land, non-native forest plantations, built environments, etc.).
4.5.2 – Population genetic structure and dispersal behaviour A generalist dispersal strategy (i.e. one where movement is not restricted to particular vegetation types or landscape attributes) is advantageous for a species which prefers ephemeral habitats that are less predictable and more variable in temporal and spatial dispersion of resources (Lee et al. 1981). For small mammals in relatively stable environments, dispersal is often restricted (Peakall et al. 2003; Kraaijeveld-Smit et al. 2007; Macqueen et al. 2008; Gardner-Santana et al. 2009). Temporal instability of habitat quality may enhance effective dispersal and/or increase dispersal propensity (Gauffre et al. 2008). For example, a dispersal strategy with movement restricted to neighbouring habitat patches would be ineffective for the eastern chestnut mouse to take advantage of recently burned heathland and access early successional stage vegetation. While there is not a clear relationship between dispersal distance and the scale of genetic spatial autocorrelation, the scale of positive autocorrelation that we identified (2–3 km; Fig. 4.4) is similar to the movement patterns of other pseudomyine mice. The smoky mouse Pseudomys fumeus and eastern pebble-mound mouse P. patrius have both been observed to make minimum movements greater than 600 m and 1 km, respectively in a single night (Ford 2003, 2008). The sandy inland mouse P. hermannsburgensis is capable of dispersing very large distances (>16 km) (Dickman et al. 1995) suggesting the eastern chestnut mouse may be physiologically capable of much longer dispersal subject to landscape constraints. Given that colonisation is limited by the dispersal-limited sex, a generalist dispersal strategy that includes an absence of sex-biased dispersal would be beneficial for a post-fire specialist species to access available high quality but unoccupied habitat. Our results show no obvious evidence for sex-bias in dispersal by the eastern chestnut mouse. This indicates that both sexes disperse or that our dataset was not large enough to discern the philopatric sex. Whether a low, or temporally inconsistent, level of sex-biased dispersal occurs and was not detectable in our genetic data is difficult to determine. However, clear genetic signals of sex- biased dispersal were detected using similar sample sizes for two other species that also occur in this landscape (or are closely-related to species in this landscape), the agile antechinus (Antechinus agilis) and the bush rat (Rattus fuscipes) (Peakall et al. 2003; Banks et al. 2005; Banks & Peakall 2012). These species are not post-fire specialists. So whatever the absolute level of dispersal by male and female eastern chestnut mice, the evidence suggests that the relative strength of sex-bias in dispersal is weaker in the eastern chestnut mouse than in other small mammals studied in this region. The lack of a signal of sex-biased dispersal (a consistent significant difference in pairwise spatial autocorrelation of male and female animals) suggests an atypical mammalian structure exists (Gardner-Santana et al. 2009). This is because male-
Pereoglou, F., Population processes in an early successional heathland 89 Chapter 4: Landscape genetics biased dispersal in mammals is considered an effective mechanism for inbreeding avoidance (Greenwood 1980; Bowler & Benton 2005). Eastern chestnut mouse dispersal behaviour may have evolved as a response to acquisition of suitable habitat rather than inbreeding avoidance, and therefore it is an advantageous strategy for both sexes to be capable of dispersal. An interesting area for future research would be to investigate whether patterns of sex-biased dispersal differ between colonisation and migration (between existing populations) dispersal events, a point raised by Wade and McCauley (1988). There is evidence that the tendency of one sex or another towards dispersal may be unpredictable and dependent on habitat quality or landscape pattern (Banks et al. 2005). For example, in the Australian bush rat Rattus fuscipes, sex-biased dispersal was absent in populations studied in south-eastern Australian forest (Kraaijeveld-Smit et al. 2007) but distinct female philopatry was found in fragmented rainforest in Queensland (Macqueen et al. 2008). Further research into eastern chestnut mouse dispersal behaviour should be replicated over multiple study areas before we are able to confirm the mechanisms postulated above.
4.5.3 – Conclusion Landscape resistance was a better predictor of genetic structure in the eastern chestnut mouse population than isolation by distance alone. Genetic connectivity was provided by burnt habitat, indicating fire may play a positive role in structuring populations of some species. Dispersal was possible through unsuitable habitat and any effect of landscape heterogeneity is likely to depend on whether heterogeneity is provided by indigenous vegetation or by anthropogenic modified landscapes. We postulate that a generalist dispersal strategy – where movement is not restricted to particular vegetation types, landscape attributes, or sex-biased – is advantageous for a species which prefers ephemeral habitat conditional on spatial and temporal variability.
4.6 Acknowledgement
We most gratefully acknowledge the Wreck Bay Aboriginal Community and the Wandandian People of the Dharawal and Dhurga language groups on whose lands this study was conducted; Christine Hayes for preliminary experimental laboratory analysis and assistance; Wendy Hartman, Alfred Wellington, Tony Carter, Geoff Kay, Steve Phillips, Marama Hopkins, Corey Green, Nick Dexter, Karen Viggers, Sally Townley, Les Mitchell and Martin Westgate for field assistance and advice; and Rod Peakall for laboratory support and comments that helped improve this manuscript. Funding was provided by the Australian Academy of Science, WV Scott Charitable Trust, Norman Wettenhall Foundation and was conducted under ANU Animal Experimentation Ethics Committee protocol C.RE.49.08, NSW National Parks and Wildlife Service Scientific Licence S12690, and Booderee National Park
Pereoglou, F., Population processes in an early successional heathland 90 Chapter 4: Landscape genetics Permit BDR08/00009. The comments of four anonymous reviewers and subject editor Lisette Waits substantially improved earlier versions of this manuscript.
4.7 References
Baguette M, Van Dyck H (2007) Landscape connectivity and animal behavior: functional grain as a key determinant for dispersal. Landscape Ecology, 22, 1117–1129. Banks SC, Dujardin M, McBurney L, et al. (2011) Starting points for small mammal population recovery after wildfire: recolonisation or residual populations? Oikos, 120, 26–37. Banks SC, Lindenmayer DB, Ward SJ, Taylor AC (2005) The effects of habitat fragmentation via forestry plantation establishment on spatial genotypic structure in the small marsupial carnivore, Antechinus agilis. Molecular Ecology, 14, 1667–1680. Banks SC, Peakall R (2012) Genetic spatial autocorrelation can readily detect sex-biased dispersal. Molecular Ecology, 21, 2092–2105. Baur A, Baur B (1992) Effect of corridor width on animal dispersal: a simulation study. Global Ecology and Biogeography Letters, 2, 52–56. Bond WJ, Keeley JE (2005) Fire as a global herbivore: the ecology and evolution of flammable ecosystems. Trends in Ecology & Evolution, 20, 387–394. Bowler DE, Benton TG (2005) Causes and consequences of animal dispersal strategies: relating individual behaviour to spatial dynamics. Biological Reviews, 80, 205–225. Bruford MW, Hanotte O, Brookfield JKY, Burke T (1988) Multilocus and single-locus DNA fingerprinting. In: Molecular Genetic Analysis of Populations: A Practical Approach (ed. Hoelzel AR), pp. 287–336. IRL Press, Oxford. Catling PC, Newsome AE (1981) Responses of the Australian vertebrate fauna to fire: an evolutionary approach. In: Fire and the Australian Biota (eds. Gill AM, Groves RH, Noble IR), pp. 273–310. Australian Academy of Science, Canberra. Clarke M. F. (2008) Catering for the needs of fauna in fire management: science or just wishful thinking? Wildlife Research, 35, 385–94. Cockburn A, Scott MP, Scotts DJ (1985) Inbreeding avoidance and male-biased natal dispersal in Antechinus spp. (Marsupialia: Dasyuridae). Animal Behaviour, 33, 908–915. Cushman SA, McKelvey KS, Hayden J, Schwartz MK (2006) Gene flow in complex landscapes: Testing multiple hypotheses with causal modeling. American Naturalist, 168, 486–499. Dickman CR, Predavec M, Downey FJ (1995) Long-range movements of small mammals in arid Australia: Implications for land management. Journal of Arid Environments, 31, 441–452. ESRI (2009) ArcMap 9.2. Environmental Systems Resource Institute, California.
Pereoglou, F., Population processes in an early successional heathland 91 Chapter 4: Landscape genetics Excoffier L, Foll M, Petit RJ (2009) Genetic Consequences of Range Expansions. Annual Review of Ecology, Evolution, and Systematics, 40, 481–501. Fahrig L (2007) Non-optimal animal movement in human-altered landscapes. Functional Ecology, 21, 1003–1015. Fahrig L, Paloheimo J (1988) Determinants of local population size in patchy habitats. Theoretical Population Biology, 34, 194–213. Flannigan MD, Krawchuk MA, de Groot WJ, Wotton BM, Gowman LM (2009) Implications of changing climate for global wildland fire. International Journal of Wildland Fire, 18, 483–507.
Foll M, Gaggiotti O (2006) Identifying the environmental factors that determine the genetic structure of Populations. Genetics 174, 875-891. Ford FD (2003) Conilurine rodent evolution: the role of ecology in modifying evolutionary consequences of environmental change, James Cook University, Townsville, Australia. Ford FD (2008) Smokey mouse, Pseudomys fumeus. In: The Mammals of Australia (eds. Van Dyck S, Strahan R), pp. 629–630. Reed New Holland, Sydney, Australia. Fox BJ (1982) Fire and mammalian secondary succession in an Australian coastal heath. Ecology, 63, 1332–1341. Fox BJ (2008) Eastern chestnut mouse, Pseudomys gracilicaudatus. In: The Mammals of Australia (eds. Van Dyck S, Strahan R), pp. 634–635. Reed New Holland, Sydney, Australia. Fox BJ, Kemper CM (1982) Growth and development of Pseudomys gracilicaudatus (Rodentia: Muridae) in the laboratory. Australian Journal of Zoology, 30, 175–185. Fox BJ, Taylor JE, Thompson PT (2003) Experimental manipulation of habitat structure: a retrogression of the small mammal succession. Journal of Animal Ecology, 72, 927–940. Franklin J, Lindenmayer D, MacMahon J (2000) Threads of continuity: ecosystem disturbances, biological legacies and ecosystem recovery. Conservation Biology in Practice, 1, 8–16. Gardner-Santana LC, Norris DE, Fornadel CM, et al. (2009) Commensal ecology, urban landscapes, and their influence on the genetic characteristics of city-dwelling Norway rats (Rattus norvegicus). Molecular Ecology, 18, 2766–2778. Gauffre B, Estoup A, Bretagnolle V, Cosson JF (2008) Spatial genetic structure of a small rodent in a heterogeneous landscape. Molecular Ecology, 17, 4619–4629. Goslee SC, Urban DL (2007) The ecodist package for dissimilarity-based analysis of ecological data. Journal of Statistical Software, 22, 1–19. Goudet J (2001) FSTAT, a program to estimate and test gene diversities and fixation indices (version 2.9.3). Available from http://www.unil.ch/izea/softwares/fstat.html.
Pereoglou, F., Population processes in an early successional heathland 92 Chapter 4: Landscape genetics Goudet J, Perrin N, Waser P (2002) Tests for sex-biased dispersal using bi-parentally inherited genetic markers. Molecular Ecology, 11, 1103–1114. Greenwood PJ (1980) Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour, 28, 1140–1162. Gros A, Hovestadt T, Poethke HJ (2008) Evolution of sex-biased dispersal: The role of sex- specific dispersal costs, demographic stochasticity, and inbreeding. Ecological Modelling, 219, 226–233. Guillot G, Mortier F, Estoup A (2005) Geneland: a computer package for landscape genetics. Molecular Ecology Notes ,5, 712–715.
Guillot, G. and F. Rousset. 2012. On the use of the simple and partial Mantel tests in presence of spatial auto-correlation. arXiv:1112.0651v1. Gustafson EJ, Gardner RH (1996) The effect of landscape heterogeneity on the probability of patch colonization. Ecology, 77, 94–107. Jaquiery J, Broquet T, Hirzel AH, Yearsley J, Perrin N (2011) Inferring landscape effects on dispersal from genetic distances: how far can we go? Molecular Ecology 20, 692-705. Kraaijeveld-Smit FJL, Lindenmayer DB, Taylor AC, MacGregor C, Wertheim B (2007) Comparative genetic structure reflects underlying life histories of three sympatric small mammal species in continuous forest of south-eastern Australia. Oikos, 116, 1819–1830.
Landguth EL, Fedy BC, Oyler-McCance SJ, et al. (2012) Effects of sample size, number of markers, and allelic richness on the detection of spatial genetic pattern. Molecular Ecology Resources, 12, 276–284. Lee AK, Baverstock PR, Watts CHS (1981) Rodents - the late invaders. In: Ecological Biogeography of Australia (ed. Keast A), pp. 1521–1553. Dr W Junk Publishers, The Hague, London. Lichstein J (2007) Multiple regression on distance matrices: A multivariate spatial analysis tool. Plant Ecology, 188, 117-131. Lindenmayer DB, Fischer J (2006) Habitat Fragmentation and Landscape Change. Island Press, Washington DC. Lindenmayer DB, MacGregor C, Welsh A, et al. (2008) Contrasting mammal responses to vegetation type and fire. Wildlife Research, 35, 395–408. Lindenmayer DB, Peakall R (2000) The Tumut experiment - integrating demographic and genetic studies to unravel fragmentation effects: a case study of the native bush rat. In: Genetics, Demography and the Viability of Fragmented Populations (eds. Young A, Clarke G), pp. 173–201. Cambridge University Press, Cambridge. Lorenz GC, Barrett GW (1990) Influence of simulated landscape corridors on house mouse (Mus musculus) dispersal. American Midland Naturalist, 123, 348–356.
Pereoglou, F., Population processes in an early successional heathland 93 Chapter 4: Landscape genetics Macqueen PE, Nicholls JA, Hazlitt SL, Goldizen AW (2008) Gene flow among native bush rat, Rattus fuscipes (Rodentia : Muridae), populations in the fragmented subtropical forests of south-east Queensland. Austral Ecology, 33, 585–593. Manel S, Schwartz MK, Luikart G, Taberlet P (2003) Landscape genetics: combining landscape ecology and population genetics. Trends in Ecology and Evolution, 18, 189–197. McRae BH, Beier P (2007) Circuit theory predicts gene flow in plant and animal populations. Proceedings of the National Academy of Sciences of the USA, 104, 19885–19890. McRae BH, Dickson BG, Keitt TH, Shah VB (2008) Using circuit theory to model connectivity in ecology, evolution and conservation. Ecology, 89, 2712–2724.
McRae BH, Shah VB (2009) Circuitscape User Guide. University of California, Santa Barbara. Available at http://www.circuitscape.org. Meek PD, Triggs B (1997) Extension to the range of the eastern chestnut mouse, Pseudomys gracilicaudatus, to Jervis Bay, New South Wales. Australian Mammalogy, 20, 123–125. Monamy V, Fox BJ (2000) Small mammal succession is determined by vegetation density rather than time elapsed since disturbance. Austral Ecology, 25, 580–587. Moro D, Spencer PBS (2003) Microsatellite primers for the western pebble-mound mouse (Pseudomys chapmani) that show cross amplification for other species of Australian rodent. Molecular Ecology Notes, 3, 259–261.
Morris DW (1992) Scales and costs of habitat selection in heterogeneous landscapes. Evolutionary Ecology, 6, 412–432. Mossman CA, Waser PM (2001) Effects of habitat fragmentation on population genetic structure in the white-footed mouse (Peromyscus leucopus). Canadian Journal of Zoology-Revue Canadienne De Zoologie, 79, 285–295. Murphy MA, Dezzani R, Pilliod DS, Storfer A (2010) Landscape genetics of high mountain frog metapopulations. Molecular Ecology, 19, 3634–3649. Park SDE (2001) Trypanotolerance in West African Cattle and the Population Genetic Effects of Selection. University of Dublin.
Pavlacky DC, Goldizen AW, Prentis PJ, Nicholls JA, Lowe AJ (2009) A landscape genetics approach for quantifying the relative influence of historic and contemporary habitat heterogeneity on the genetic connectivity of a rainforest bird. Molecular Ecology, 18, 2945–2960. Peakall R, Ruibal M, Lindenmayer DB (2003) Spatial autocorrelation analysis offers new insights into gene flow in the Australian bush rat, Rattus fuscipes. Evolution, 57, 1182– 1195. Peakall R, Smouse PE (2006) GenAlEx 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes, 6, 288–295.
Pereoglou, F., Population processes in an early successional heathland 94 Chapter 4: Landscape genetics Pereoglou F, Macgregor C, Banks SC, et al. (2011) Refuge site selection by the eastern chestnut mouse in recently burnt heath. Wildlife Research, 38, 290–298. Perez-Espona S, Perez-Barberia FJ, McLeod JE, et al. (2008) Landscape features affect gene flow of Scottish Highland red deer (Cervus elaphus). Molecular Ecology, 17, 981–996. Pritchard JK, Stephens M, Donnelly P (2000) Inference of population structure using multilocus genotype data. Genetics, 155, 945–959. R Development Core Team (2008) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Ruscoe WA, Mather PB, Wilson J (1998) Genetic structure in populations of Rattus sordidus in sugarcane–growing districts of Queensland, Australia. Journal of Mammalogy, 79, 612– 623. Schrey AW, Ashton KG, Heath S, McCoy ED, Mushinsky HR (2011) Fire Alters Patterns of Genetic Diversity Among 3 Lizard Species in Florida Scrub Habitat. Journal of Heredity, 102, 399–408. Schrey AW, Fox AM, Mushinsky HR, McCoy ED (2010) Fire increases variance in genetic characteristics of Florida Sand Skink (Plestiodon reynoldsi) local populations. Molecular Ecology, 20, 56–66. Schuelke M (2000) An economic method for the fluorescent labeling of PCR fragments. Nature Biotechology, 18, 233–234. Schultz CB (1998) Dispersal behavior and its implications for reserve design in a rare oregon butterfly. Conservation Biology, 12, 284–292. Scribner KT, Blanchong JA, Bruggeman DJ, et al. (2005) Geographical genetics: conceptual foundations and empirical applications of spatial genetic data in wildlife management. Journal of Wildlife Management, 69, 1434–1453. Selonen V, Hanski IK (2004) Young flying squirrels (Pteromys volans) dispersing in fragmented forests. Behavioral Ecology, 15, 564–571. Short Bull RA, Cushman SA, Mace R, et al. (2011) Why replication is important in landscape genetics: American black bear in the Rocky Mountains. Molecular Ecology, 20, 1092– 1107. Smouse PE, Peakall R (1999) Spatial autocorrelation analysis of individual multiallele and multilocus genetic structure. Heredity, 82, 561–573. Smouse PE, Peakall R, Gonzales E (2008) A heterogeneity test for fine-scale genetic structure. Molecular Ecology, 17, 3389–3400. Spear SF, Balkenhol N, Fortin M-J, McRae BH, Scribner KIM (2010) Use of resistance surfaces for landscape genetic studies: considerations for parameterization and analysis. Molecular Ecology, 19, 3576–3591.
Pereoglou, F., Population processes in an early successional heathland 95 Chapter 4: Landscape genetics Spear SF, Peterson CR, Matocq MD, Storfer A (2005) Landscape genetics of the blotched tiger salamander (Ambystoma tigrinum melanostictum). Molecular Ecology, 14, 2553–2564. Spear SF, Storfer A (2008) Landscape genetic structure of coastal tailed frogs (Ascaphus truei) in protected vs. managed forests. Molecular Ecology, 17, 4642–4656. Spear SF, Storfer A (2010) Anthropogenic and natural disturbance lead to differing patterns of gene flow in the Rocky Mountain tailed frog, Ascaphus montanus. Biological Conservation, 143, 778–786. Steffen W, Burbidge AA, Hughes L, et al. (2009) Australia's Biodiversity and Climate Change CSIRO Publishing, Melbourne.
Storfer A, Murphy MA, Spear SF, Holderegger R, Waits LP (2010) Landscape genetics: where are we now? Molecular Ecology, 19, 3496–3514. Stow AJ, Magnusson WE (2012) Genetically defining populations is of limited use for evaluating and managing human impacts on gene flow. Wildlife Research, 39, 290–294. Taws N (1997) Vegetation survey and mapping of Jervis Bay Territory. Report. Environment Australia, Canberra. Taylor PD, Fahrig L, Henein K, Merriam G (1993) Connectivity is a vital element of landscape structure. Oikos, 68, 571–573. Templeton AR, Brazeal H, Neuwald JL (2011) The transition from isolated patches to a metapopulation in the eastern collared lizard in response to prescribed fires. Ecology, 92, 1736–1747. Templeton AR, Robertson RJ, Brisson J, Strasburg J (2001) Disrupting evolutionary processes: the effect of habitat fragmentation on collared lizards in the Missouri Ozarks. Proceedings of the National Academy of Sciences of the USA, 98, 5426–5432. Turner MG, Baker WL, Peterson CJ, Peet RK (1998) Factors influencing succession: lessons from large, infrequent natural disturbances. Ecosystems, 1, 511–523. Wade MJ, McCauley DE (1988) Extinction and Recolonization: Their Effects on the Genetic Differentiation of Local Populations. Evolution, 42, 995–1005.
Wang Y-H, Yang K-C, Bridgman C, Lin L-K (2008) Habitat suitability modeling to correlate gene flow with landscape connectivity. Landscape Ecology, 23, 989–1000. Watts CHS, Kemper CM (1989) Muridae. In: Fauna of Australia, Mammalia. (eds. Walton DW, Richardson BJ), pp. 939–956. AGPS, Canberra. Williams AAJ, Karoly DJ, Tapper N (2001) The sensitivity of Australian fire danger to climate change. Climatic Change, 49, 171–191. Wright S (1943) Isolation by distance. Genetics, 28, 139–156.
Pereoglou, F., Population processes in an early successional heathland 96
Chapter 5: Population dynamics
Population dynamics
Submitted as:
Pereoglou, F., Banks, S.C., MacGregor, C., Ford, F., Wood, J., Lindenmayer, D.B. How does fire history influence demographic parameters and genetic patterns in an early successional specialist small mammal? Ecology and Evolution, Submitted
Pereoglou, F., Population processes in an early successional heathland 97 Chapter 5: Population dynamics
5.1 Abstract Understanding how demographic parameters of post-fire specialists vary with fire history can provide insights into the population viability and distribution of such species, including their response to landscape management and novel fire regimes. We used an innovative combination of demographic and genetic analyses to quantify survival, recruitment and immigration rates in populations of an early successional specialist small mammal, the Australian eastern chestnut mouse (Pseudomys gracilicaudatus). Our aim was to understand how key demographic parameters vary across the landscape in response to fire history related environmental factors. Estimates of survival and recruitment varied across sites. Survival was influenced by fire history and elements of habitat structure affected by fire. Survival was highest among adults with recruitment driven predominantly by juveniles entering the population prior to the breeding season. Genetic assignment tests revealed immigration contributed little to recruitment compared to reproduction, suggesting populations are demographically discrete at this post-fire stage. This was concordant with findings of significant genetic differentiation between sites. Demographic and genetic data were consistent; sites with high recruitment also had higher genetic diversity and lower temporal genetic structure, consistent with reduced genetic drift. In light of landscape genetics analyses showing that fire promotes population connectivity, this study suggests a two-phase model of spatial population dynamics, where post- fire colonization is followed by locally driven population dynamics, in which survival rates are influenced by fire-history effects on habitat quality. Thus, post-fire specialists can be dependent on landscape-level connectivity and site-level habitat quality, both of which are influenced by fire history.
5.2 Introduction
Why do populations of successional specialists show such strong associations with disturbance history? Many early successional specialist species are sensitive to changes in land management and disturbance regimes. For post-fire specialists, understanding how fire history influences population dynamics through key demographic processes like survival, recruitment and migration is important for understanding the viability and distribution of such species in fire-prone landscapes. Site- and landscape-level processes drive environmental variation and thus are expected influences on these demographic rates. Estimating survival and recruitment is a basic requirement for identifying factors that influence the rate of population change and limit population viability (Sandercock, 2006, Lebreton et al., 1992, Mills, 2013). Immigration and emigration also contribute to the change of population size in open populations (Schaub et al., 2013), with the exchange of individuals among populations affecting their dynamics and persistence (Schaub et al., 2012). The rate of immigration is notoriously difficult to estimate and
Pereoglou, F., Population processes in an early successional heathland 98 Chapter 5: Population dynamics consequently relatively little is known about its role in population dynamics (Abadi et al., 2010). Innovative combinations of demographic and genetic data can overcome this problem, producing estimates of survival and recruitment in addition to evaluating the relative contributions of reproduction and immigration to recruitment. Variation in ecological conditions generated by disturbance can alter demographic and genetic patterns (Banks et al., 2013). For instance, fire influences habitat suitability for many species, and populations of species that specialize in fire-related successional conditions can occur in transient, or ephemeral, patches of habitat that are defined by fire history. Understanding fundamental population processes such as recruitment, immigration and survival, and how these respond to disturbance history will be important for predicting and managing the impacts of changing disturbance regimes on the persistence of successional specialist species. Studies on a number of species have shown that fire history influences dispersal by early successional species, with consequences for genetic structure (Pereoglou et al., 2013, Templeton et al., 2011, Pierson et al., 2013). Genetic variation can be lost by increasing genetic drift as vegetation regeneration progresses and late successional communities develop, impacting the quality, spatial configuration and connectivity of habitat (Templeton et al., 2001). Reduced habitat quality and connectivity for early successional species as a consequence of succession can result in associated population declines of early successional fauna (Betts et al., 2010). The integration of genetic and demographic data will enable us to understand how population processes of such species are impacted by environmental variation, accounting for the effects of survival, breeding success and animal movements in an open population by identifying the origin of recruits (Mills, 2013). Measuring demographic rates can be difficult as they vary with individual characteristics and fluctuate in response to environmental variability (Lima et al., 2001, Villella et al., 2004, Boyce et al., 2006, Ives, 2009). Individual-level drivers are those regulated by internal processes and include sex and age, which have been associated with survival, movement and recruitment in many species (Aars and Ims, 2002, Holland and Bennett, 2010, Pryde et al., 2005, Troyer et al., 2014, Sibly et al., 1997). External processes are those occurring at the site- and landscape- levels. For species with fire-related successional habitat specialization, site-level drivers of population dynamics relate to fire history or other fixed, geographically driven environmental variables relating to habitat quality. For early successional species, fire can generate resources that are critical for survival and reproductive success (Hutto, 1995, Templeton et al., 2011). Landscape-level drivers of population dynamics include patch size, landscape heterogeneity, and structural connectivity of habitat. Landscape structure influences animal movements (i.e. dispersal, immigration), which subsequently influence colonization, reproduction and genetic structure (Hanski, 2009, Donner et al., 2010). Survivorship has been often been positively associated with patch size (Barbour and Litvaitis, 1993, Bowers et al., 1996, Villafuerte et al., 1997). However, the effect of landscape heterogeneity, or fragmentation, on key demographic
Pereoglou, F., Population processes in an early successional heathland 99 Chapter 5: Population dynamics parameters appears to be much more complicated and inconsistent (see also Umapathy and Kumar, 2000, Dooley and Bowers, 1998, Nupp and Swihart, 1996, Soga and Koike, 2013). To understand how population processes respond to environmental variation, we need to evaluate internal drivers on key demographic rates like survival, recruitment and immigration, and investigate site- and landscape-level factors as external drivers of those processes. In this study, we used a combination of demographic mark-recapture analyses and microsatellite genetic data to address questions about drivers of demographic processes. We used mark-recapture and genetic analyses to test hypotheses about effects of internal ‘state’ variables (such as sex and age), external site-level factors (such as fire history and fire-driven habitat variables) and landscape-level factors (such as patch size and heterogeneity) on survival. We also tested how these factors influence recruitment, separating the contributions of immigration and reproduction to recruitment with analyses of microsatellite genetic data. Integrating demographic and genetic analyses has benefits in both directions, improving our understanding of demographic population processes, but also improving our understanding of the demographic drivers of spatial and temporal variation in genetic diversity and structure. We show that species with genetically structured populations have demographic processes influenced more strongly by site-level factors than landscape-level factors, because populations that are discrete are likely to be strongly influenced by local resources.
5.3 Materials and methods
5.3.1 Study species
The eastern chestnut mouse is an early successional specialist that occurs in heathland regenerating after fire (Fox, 1982). Distributed in disjunct populations along the east coast of Australia, the species is considered extinct in many parts of its range (Breed and Ford, 2007) and is currently listed as Vulnerable on Schedule 2 of the Threatened Species Conservation Act 1995 (NSW) (Available at http://www.austlii.edu.au/au/legis/nsw/consol_act/tsca1995323/). Breeding occurs annually during the summer months when adult females can produce multiple litters of one to five (usually three) offspring (Fox, 2008). Adults show significant sexual dimorphism (Fox and Kemper, 1982) and attain average weights of 90 g for males and 75 g for females (Fox, 2008). Diurnal refuge sites are associated with the presence of specific structures including grass trees (Xanthorrhea spp.), buttongrass (Gymnoschoenus sp.), collapsed dead stems, and areas of tall, dense vegetation (Pereoglou et al., 2011). Individuals are transient and non-gregarious in their use of refugia (Pereoglou et al. 2011). Little is known about social organization and behavior. Radio-tracking data suggests that adults are solitary and territorial, with females having a home range of 0.5 – 1 ha and up to 4 ha for males (Pereoglou et al. unpublished data).
Pereoglou, F., Population processes in an early successional heathland 100 Chapter 5: Population dynamics 5.3.2 Study area
Booderee National Park is located approximately 200 km south of Sydney, Australia (Fig. 5.1). Six types of vegetation occur in the Park consisting of forest, woodland, rainforest, heathland, sedgeland, and shrubland (Taws, 1997). We selected seven sites in heathland for a two-year mark-recapture study. In heathland sites, vegetation was generally less than 2 m tall and varyingly dominated by Banksia ericifolia and Sprengelia incarnata, with Allocasuarina distyla, Isopogon anemonifolius and Hakea teretifolia also being common.
Figure 5.1: Map of Booderee national park indicating study sites and vegetation types.
5.3.3 Animal capture and marking We captured animals using grids of aluminium box traps (Elliott Scientific Equipment, Australia) open for a minimum of four consecutive nights during each survey season: November 2008 – March 2009 (S89), June – August 2009 (W9), November 2009 – March 2010 (S910), and June – August 2010 (W10). Grid arrangement comprised lines at 20 m intervals, with traps set 10 m apart, and with sites covering an area of 1 – 2.5 hectares. We baited traps with rolled oats, peanut butter and fennel seed, checked them daily and replaced traps that captured a non-target species. During winter, we insulated traps with non-absorbent cotton wool and a plastic sleeve. We marked captures using Trovan ID100 implantable transponders (Microchips Australia Ltd, Australia), assessed species, sex and age (breeder, B or non-breeder,
Pereoglou, F., Population processes in an early successional heathland 101 Chapter 5: Population dynamics NB), and collected a 2 mm ear biopsy that was stored in 100% ethanol. All required ethics guidelines were followed. Our research was conducted under The Australian National University Animal Experimentation Ethics Committee protocol C.RE.49.08 and Booderee National Park Permit BDR08/00009.
5.3.4 Genetic data analysis
We genotyped individuals (n = 223) at nine microsatellite loci. DNA extraction, PCR amplification and genotyping protocols were undertaken as described in Pereoglou et al. (2013). To identify identical genotypes and putative recaptured individuals in different trapping seasons (to account for potential microchip loss), we used the genotype matching function of MSTOOLS (Park, 2001). Duplicate genotypes were re-labelled if the corresponding capture records were consistent with multiple captures and samples from the same individual (e.g. due to temporary marking or tag loss). We pooled genotypes from all survey seasons to calculate standard population genetic summary statistics for all loci and sites including the number of alleles (Na), observed (HO) and expected (HE) heterozygosities, and fixation index (F). We tested for departures from Hardy-Weinberg equilibrium using FSTAT (Goudet, 2001). We estimated population subdivision (overall FST), temporal genetic differentiation between years
(tFST) at each site, and pairwise FST between sites. To test for isolation by distance on pairwise
FST between sites (FST/1-FST versus geographic distance), we used a Mantel test. We estimated population-specific genetic differentiation (pFST) using GESTE 2.0 (Foll and Gaggiotti, 2006). Unless otherwise stated, we performed all permutational tests 999 times and analyses were completed in GENALEX 6.3 (Peakall and Smouse, 2006). To assess annual immigration at each site, we pooled genotype data from survey seasons in each year. We used the assignment or exclusion of individuals test in GENECLASS 2.0 (Piry et al., 2004) to compute the probability that an individual belonged to its sample population. Individuals were simulated 100 times using the Rannala and Mountain (1997) criteria for computation, the Monte-Carlo re-sampling algorithm described in Paetkau et al. (2004), and assignment probability thresholds of 0.05 and 0.01.
5.3.5 Estimating survival and recruitment
For each individual, we constructed an encounter history consisting of a contiguous series of 1’s and 0’s representing each survey season, where ‘1’ indicates that an animal was recaptured (or otherwise known to be alive and in the sampling area), and ‘0’ indicates the animal was not recaptured (or otherwise seen) (Cooch and White, 2014). We used Program MARK 7.1 (White and Burnham, 1999) to model factors influencing variation in seasonal survival and recruitment. We estimated survival using the standard Cormack-Jolly-Seber (CJS) model based on live animal captures in an open population (Lebreton et al., 1992), and survival and recruitment using Pradel models (Pradel, 1996a). We used terminology as defined by Pereoglou, F., Population processes in an early successional heathland 102 Chapter 5: Population dynamics
Cooch and White (2014): ‘apparent survival’ (phi or ), ‘recapture probability’ (p) and ‘recruitment probability’ (f). Apparent survival is the probability of remaining in the study population. Recapture probability is the probability that an animal alive in the study population at the time of survey will be captured. Recruitment probability is the number of individuals entering the population between timei and timei + 1 per individual already in the population at timei. We categorised data into seven groups based on the site of capture. Because the length of inter-session periods varied, we set time intervals to the number of days, averaged across the seven sites. To test for departure of data from the underlying assumptions of the CJS model, we used the global goodness-of-fit method available in the program U-CARE version 2.0 (Choquet et al., 2009). The general model was an adequate fit for our data (GOF test df=13, x2=11.591, p=0.561) and no adjustment to subsequent model metrics was necessary because the variance inflation factor (c-hat), obtained by dividing the model x2 statistic by the degrees of freedom, was < 1. We analysed data for sex, age, site and season effects, with a set of covariates describing habitat, landscape features and genetic composition of local populations (= site). We sampled vegetation in two 20 x 20 m plots at each site, and averaged values for the number of dead shrubs present (DEAD SHRUB), the number of understorey species present (FLORISTIC RICHNESS), and the understorey (< 2 m) percent cover of all plant species combined (COVER) across the two vegetation plots. We measured landscape features from digital layers of vegetation mapping by Taws (1997) and spatially-explicit fire mapping conducted by Booderee National Park in ArcGIS 9.2 (ESRI, 2009). For each site, we recorded the number of fires that had occurred within a 100 m radius of each site since fire mapping began in the mid- 1950s (FIRES), and the contiguous area of the heathland habitat patch within which each site was located (AREA) were recorded. As a measure of landscape heterogeneity, we counted the number of vegetation types within a 500 m radius of each site (LANDSCAPE). We calculated the genetic composition of local populations (=sites) according to the analytical methods in the preceding section and included genetic diversity (observed heterozygosity, HO), population- specific genetic differentiation (pFST), and temporal FST within each site (tFST). We fit data to CJS models that tested the main effects of site and individual-state variables (AGE, SEX) over time (SEASON). We then constructed models with fire (FIRES), habitat
(DEAD SHRUB, FLORISTIC RICHNESS, COVER) and landscape-level (AREA, LANDSCAPE) variables to evaluate major influences of survival variation. Models were run with a constant recapture probability. We also modeled survival and recruitment probabilities (Pradel, 1996b) to explore relationships between individual-state, habitat, landscape-level variables, and genetic variability (HO, pFST, tFST) on recruitment patterns. While fire, habitat and landscape-level variables potentially influence recruitment variation among sites, the genetic site variables respond to the recruitment patterns (e.g. was recruitment rate related to genetic drift), so we expected to learn something different by evaluating these. Following on from the
Pereoglou, F., Population processes in an early successional heathland 103 Chapter 5: Population dynamics CJS models, we set Pradel models to have both site-dependent survival rates and a constant recapture probability. Model selection was based on Akaike’s information criterion corrected for small sample size bias (AICc, Burnham and Anderson, 2002, Cooch and White, 2014). The difference between AICc for each model (i, AICc difference) and that for the model with the smallest observed AICc from the set of models considered was calculated along with the Akaike weight for each model (wi). As a guide i < 2 should be considered to have substantial support and used for making inferences, models with i of 4-7 have considerably less support, and models with i > 10 essentially have no support (Burnham and Anderson, 2002). Model averaging of parameter estimates and their unconditional standard errors were calculated according to Burnham and Anderson (2002).
5.4 Results
Seasonal encounters (n = 278) represented 181 individuals of the eastern chestnut mouse from the seven populations (Table 5.1). No recaptures were recorded outside of the population of their initial capture.
5.4.1 Population genetics summary statistics
Departures from Hardy-Weinberg expectations were not beyond the false error rate for multiple testing (i.e. seven out of 130 tests were likely to be significant), nor were they consistent across sites or years (e.g. at locus 7D12, sites 7-44 and 7-P in 2008-09, and site 7-46 in 2009-10). We identified between two and five alleles per locus. Expected heterozygosity across sites ranged from 0.463 – 0.644, with observed heterozygosity greater than expected in all sites except 7-44 (Table 5.1).
5.4.2 Population subdivision and spatial genetic structure
We identified significant genetic differentiation among sites across the study area (overall
FST = 0.170, p = 0.001) (Table 5.2) and an effect of isolation by distance. Euclidean distance
(geographic distance) was significantly correlated with pairwise FST between sites (Mantel r =
0.766, p = 0.004). Population-specific genetic differentiation (pFST) ranged from 0.047 – 0.207 and temporal (among subsequent years) FST within each site (tFST) ranged from 0 – 0.168 (Table 5.1).
5.4.3 Assignment tests
Immigration appeared to be a negligible source of recruitment. When data were pooled over all seasons, we excluded only three individuals (out of a total of 223 individuals analysed) from the populations in which they were sampled. When the genetic reference populations for
Pereoglou, F., Population processes in an early successional heathland 104 Chapter 5: Population dynamics
Table 5.1: Eastern chestnut mouse capture data, population genetics summary statistics, and assignment tests by site. E = seasonal encounters, nTC = total individual captures, M = male, F = female, pFST = population specific genetic differentiation, tFST = temporal FST within each site, nTG = total genotyped samples, Na = average number of alleles, HO = observed heterozygosity, HE = expected heterozygosity, F = fixation index, nA89 = number of genotypes assigned to populations in 2008-09, nA910 = number of genotypes assigned to populations in 2009-10, A0.05 = number of individuals that were excluded from their sampled population at p = 0.05, A0.01 = number of individuals that were excluded from their sampled population at p = 0.01
Data pooled 2008-09 2009-10 Site E nTC M F A0.05 A0.01 pFST tFST nTG Na s.e. HO s.e. HE s.e. F s.e. nA89 A0.05 A0.01 nA910 A0.05 A0.01 7-2 48 29 13 16 0 0 0.100 0.020 38 3.33 0.24 0.53 0.07 0.50 0.06 -0.06 0.03 23 0 0 15 0 0 7-44 8 6 1 5 0 0 0.097 10 2.89 0.20 0.57 0.09 0.59 0.03 0.06 0.12 8 0 0 2 7-44B 53 28 13 15 1 0 0.103 0.000 31 3.44 0.24 0.59 0.03 0.58 0.03 -0.03 0.04 21 0 0 10 0 0 7-46 62 48 22 26 1 0 0.047 0.021 58 3.78 0.28 0.60 0.06 0.57 0.05 -0.06 0.03 38 0 0 20 1 0 7-H 49 36 17 19 0 0 0.207 0.012 41 3.00 0.29 0.46 0.05 0.46 0.05 -0.03 0.03 23 0 0 18 0 0 7-I 37 22 10 12 1 0 0.106 0.168 28 3.33 0.17 0.54 0.05 0.55 0.03 0.03 0.06 19 0 0 9 0 0 7-P 21 12 5 7 0 0 0.063 0.000 17 3.11 0.20 0.64 0.06 0.55 0.04 -0.16 0.07 12 0 0 5 0 0 Total 278 181 81 100 3 0 223 144 0 0 79 1 0
Pereoglou, F., Population processes in an early successional heathland 105 Chapter 5: Population dynamics
Table 5.2: Pairwise site FST for the eastern chestnut mouse. FST values below diagonal. Probability values based on 999 permutations are shown above diagonal. All survey years and loci are combined. Non- significant values highlighted in bold.
7-2 7-44 7-44B 7-46 7-H 7-I 7-P 7-2 7-44 7-44B 7-46 7-H 7-I 7-P 7-2 0 0.001 0.001 0.001 0.001 0.001 0.022 7-2 0 0.001 0.001 0.001 0.001 0.001 0.022 7-44 0.232 0 0.001 0.001 0.001 0.003 0.001 7-44 0.232 0 0.001 0.001 0.001 0.003 0.001 7-44B 0.152 0.115 0 0.001 0.001 0.001 0.001 7-44B 0.152 0.115 0 0.001 0.001 0.001 0.001 7-46 0.073 0.142 0.147 0 0.001 0.001 7-46 0.073 0.142 0.147 0 0.001 0.001 0.347 7-7H- H 0.2870.287 0.2540.254 0.2420.242 0.2300.230 00 0.001 0.001 7-7I -I 0.1900.190 0.1140.114 0.1530.153 0.1350.135 0.2710.271 0 0.001 7-7P- P 0.0400.040 0.1320.132 0.1000.100 0.0030.003 0.210.2133 0.107 00
assignment were split by year, the rate of exclusion from the site of sampling was even lower: no individuals could be excluded from their sampling localities in 2008-09, and only one individual was excluded from its sampling locality in 2009-10 (Table 5.1).
5.4.4 Survival (CJS) models
Model averaged estimates of apparent survival were between 0.28 – 0.53 depending on the site, and recapture probability was 0.97 0.03 (Table 5.3). All models described apparent survival better than the global model (model 26 i =11.34, Table 5.4). Modelled unconstrained site variation (model 10 i =5.28) ranked higher than the constant-only model (model 19 i =7.86) indicating survival varied between sites (Table 5.4). The time constant age model (model
15 i =6.82) also ranked higher than the constant-only model. The survival rate of adults was greater than that of non-breeders ((AGE) ± S.E. = -0.47 ± 0.27, 95% CI = -1.00 – 0.05; breeder = 0.47 ± 0.05, non-breeder = 0.36 ± 0.04). Support for models with age effects that varied over time was limited (models 21 i =8.89 and 25 =i 10.52). Further, there was little support for a model with an effect of sex on survival (model 24 i =9.90). The modelled sex effects measured by the beta values suggest there was no meaningful difference in survival rates between males and females ((SEX) ± S.E. = -0.02 ± 0.27, 95% CI = -0.55 – 0.50; male = 0.41 ± 0.05, female = 0.42 ± 0.04). Overall, models with specific fire and habitat covariates had the greatest model weights. The combined effects of habitat features (dead shrubs, floristic richness and cover; models 1 – 7, Table 5.4) were clearly the strongest influence on survival. Models that included dead shrubs as a survival covariate had substantial support (Table 5.4); survival was greatest in sites with low dead shrub density ((DEADSHRUB) ± S.E. = -5.30 ± 1.83, 95% CI = -8.89 – -1.71). Model combinations of fire and habitat features were also supported (Table 5.4). The main effect of fires indicated an increased incidence of fire had a positive effect on survival
((FIRES) ± S.E. = 0.23 ± 0.10, 95% CI = 0.04 – 0.41). There was less support for survival models that included landscape heterogeneity (model 12 i =5.43) and essentially no support for models that included patch area (model 22 i =9.54). Pereoglou, F., Population processes in an early successional heathland 106
Chapter 5: Population dynamics
Table 5.3: Cormack-Jolly-Seber (CJS) and Pradel model averaged parameter estimates presented with unconditional standard error and 95 % confidence interval in parentheses; = apparent survival, p = recapture probability, f1-3 = recruitment probability (for 1st, 2nd, and 3rd time interval).
Site S.E. p S.E. f 1 S.E. f 2 S.E. f 3 S.E. CJS estimates 7-2 0.490.07 (0.35-0.64) 0.970.03 (0.80-1.00) 7-44 0.440.09 (0.28-0.61) 0.970.03 (0.80-1.00) 7-44B 0.520.07 (0.39-0.65) 0.970.03 (0.80-1.00) 7-46 0.280.06 (0.18-0.43) 0.970.03 (0.80-1.00) 7-H 0.330.08 (0.20-0.49) 0.970.03 (0.80-1.00) 7-I 0.450.06 (0.33-0.57) 0.970.03 (0.80-1.00) 7-P 0.530.10 (0.34-0.71) 0.970.03 (0.80-1.00) Pradel estimates 7-2 0.440.07 (0.31-0.57) 0.980.02 (0.89-1.00) 0.110.31 (0.00-0.98) 3.665.74 (-7.59-14.90) 0.010.01 (-0.02-0.03) 7-44 0.290.15 (0.09-0.63) 0.980.02 (0.89-1.00) 0.130.37 (0.00-0.99) 4.297.16 (-9.73-18.32) 0.010.02 (-0.03-0.04) 7-44B 0.520.11 (0.32-0.72) 0.980.02 (0.89-1.00) 0.120.33 (0.00-0.98) 4.096.56 (-8.78-16.95) 0.010.02 (-0.03-0.04) 7-46 0.320.09 (0.18-0.51) 0.980.02 (0.89-1.00) 0.140.37 (0.00-0.99) 4.477.34 (-9.90-18.85) 0.010.02 (-0.03-0.04) 7-H 0.360.07 (0.24-0.51) 0.980.02 (0.89-1.00) 0.090.27 (0.00-0.98) 2.894.68 (-6.29-12.06) 0.000.01 (-0.02-0.03) 7-I 0.440.07 (0.30-0.59) 0.980.02 (0.89-1.00) 0.110.31 (0.00-0.98) 3.675.85 (-7.79-15.13) 0.010.01 (-0.02-0.03) 7-P 0.470.11 (0.27-0.68) 0.980.02 (0.89-1.00) 0.160.63 (0.00-1.00) 5.3111.34 (-16.93-27.54) 0.010.02 (-0.04-0.05)
Pereoglou, F., Population processes in an early successional heathland 107 Chapter 5: Population dynamics
Table 5.4: Model summary describing factors affecting survival of the eastern chestnut mouse: Akaike’s information criterion for small sample size (AICc), AICc difference (i), AICc model weight (wi), parameters (K), = survival; p = recapture probability; sex = male, female; age = breeder or non-breeder; season; site; covariates in parentheses. 1
Model 2 AICc i wi K 1 (dead shrubs+floristic3 richness)p(.) 329.31 0.00 0.141 4 2 (dead shrubs)p(.)4 330.50 1.19 0.078 3 3 (dead shrubs+floristic5 richness+fires)p(.) 330.66 1.35 0.072 5 4 (dead shrubs+cover)p(.)6 331.04 1.73 0.059 4 5 (dead shrubs+floristic7 richness+cover)p(.) 331.36 2.05 0.051 5 6 (dead shrubs+fires)p(.)8 331.41 2.10 0.049 4 7 (dead shrubs+floristic9 richness+fires+cover)p(.) 332.68 3.37 0.026 6 8 (fires)p(.)10 333.41 4.10 0.018 3 9 (fires+cover)p(.)11 333.86 4.55 0.015 4 10 (site)p(.)12 334.59 5.28 0.010 8 11 (dead shrubs+floristic13 richness+fires+cover+landscape+area)p(.) 334.59 5.28 0.010 8 12 (landscape)p(.)14 334.74 5.43 0.009 3 13 (fires+floristic15 richness)p(.) 335.06 5.75 0.008 4 14 (cover)p(.)16 335.90 6.59 0.005 3 15 (age)p(.)17 336.12 6.82 0.005 3 16 (landscape+area)p(.)18 336.53 7.22 0.004 4 17 (site+season19 )p(.) 336.61 7.31 0.004 10 18 (floristic20 richness+cover)p(.) 337.16 7.85 0.003 4 19 (.)p(.)} 21 337.17 7.86 0.003 2 20 (season)p(.)22 337.95 8.64 0.002 4 21 (age+season23 )p(.) 338.20 8.89 0.002 5 22 (area)p(.)24 338.84 9.54 0.001 3 23 (floristic25 richness)p(.) 339.07 9.76 0.001 3 24 (sex)p(.)26 339.21 9.90 0.001 3 25 (age*season)p(.) 339.82 10.52 0.001 7 26 (site+season+sex+age)p(site+season+sex+age) 340.65 11.34 0.000 19
Pereoglou, F., Population processes in an early successional heathland 108 Chapter 5: Population dynamics
5.4.5 Recruitment (Pradel) models
To estimate recruitment, we modelled apparent survival with unconstrained site effects because the variable ‘site’ was expected to capture all the variation attributed to significant site-level covariates revealed by the CJS modelling. Model averaged estimates of apparent survival were between 0.29 and 0.52, and recapture probability was 0.98 0.02 (Table 5.3). Recruitment was highest in the winter 2009 (W9) to summer 2009 – 2010 (S910) interval and lowest for the summer 2009 – 2010 (S910) – winter 2010 (W10) interval (range 0 – 5.31, Table 5.3). The best supported models of per capita recruitment included age and season and population genetic parameters (HO, pFST, tFST) (models 1 – 4, Table 5.5). We attributed per capita recruitment to the in situ production of offspring, which was highest between winter 2009 (W9) and summer
2009 – 2010 (S910) (non-breeder ft1 S.E. = 0.36 0.09, ft2 = 11.04 3.86, ft3 = 0.02 0.02; breeder ft1 S.E. = 1.71-09 5.32-07, ft2 = 5.19-08 1.62-05, ft3 = 8.17-11 2.55-08).
Recruitment rate increased with overall genetic diversity ((HO) ± S.E. = 5.50 ± 3.45, 95% CI =
-1.27 – 12.27), and decreased with population specific FST ((pFST) ± S.E. = -4.18 ± 3.27, 95%
CI = -10.58 – 2.22) and temporal FST within each site ((tFST) ± S.E. = -0.47 ± 3.97, 95%CI = - 8.24 – 7.31) (Table 5.5). When we included site variation in the best Pradel models (models 5 – 8, Table 5.5), the estimates derived from the models changed but there was no difference in AICc weight between the models indicating any combination of genetic covariates with age, season, and site had an equal influence on recruitment. Models that included site-level covariates were unsupported (Table 5.5).
5.5 Discussion
Integrating demographic and genetic analyses improves both our understanding of demographic population processes, and the underlying drivers of spatiotemporal variation in genetic diversity and structure. Our demographic analysis indicated survival rates were influenced by external factors including an increased incidence of fire and specific resources affected by fire (collapsed dead shrub layer, floristic richness, and cover). Recruitment was more strongly explained by site variation than landscape structure, which is consistent with low connectivity between populations, supporting the inference of discrete populations from FST. Site-level fire history is therefore a key factor driving demographic processes in these populations over the period surveyed. Given immigration was rare, apparent survival as predicted in our modelling is likely to be a good estimate of actual survival, where individuals that disappear from the population are attributed to mortality not and emigration. The addition of genetic data improved our demographic inference by revealing reproduction by resident animals was a greater contributor to recruitment than immigration and that the strong genetic structure identified, confirmed populations are largely discrete at the post-fire period of
Pereoglou, F., Population processes in an early successional heathland 109 Chapter 5: Population dynamics
Table 5.5: Model summary describing factors affecting eastern chestnut mouse recruitment: Akaike’s information criterion for small sample size (AICc), AICc difference (i), AICc model weight (wi), parameters (K), = apparent survival; p = recapture probability; f = recruitment probability; age = breeder or non-breeder; season; site; HO = observed heterozygosity; pFST = population specific genetic differentiation; tFST = temporal FST; covariates in parentheses.
Model AICc i wi K
1 (site)p(.)f1(age+ season+HO) 510.46 0.00 0.076 12 2 (site)p(.)f2(age+ season) 510.97 0.51 0.059 11 3 (site)p(.)f3(age+ season+pFST) 511.45 0.99 0.046 12 4 (site)p(.)f4(age+ season+tFST) 513.14 2.68 0.020 12 5 (site)p(.)f5(age+ season+site) 516.69 6.23 0.003 17 6 (site)p(.)f6(age+ season+site+HO) 516.69 6.23 0.003 17 7 (site)p(.)f7(age+ season+site+pFST) 516.69 6.23 0.003 17 8 (site)p(.)f8(age+ season+site+tFST) 516.69 6.23 0.003 17 9 (site)p(.)f9(age) 624.13 113.67 0.000 9 10 (site)p(.)10f(age+H O) 628.23 117.77 0.000 11 11 (site)p(.)11f(age+ pFST) 628.40 117.94 0.000 11 12 (site)p(.)12f(age+ tFST) 628.45 117.99 0.000 11 13 (site)p(.)13f(age+site) 633.36 122.90 0.000 15 14 (site)p(.)14f(season ) 703.18 192.73 0.000 11 15 (site)p(.)15f(site+ season) 711.62 201.16 0.000 17 16 (site)p(.)16f(deadshrub) 765.90 255.44 0.000 10 17 (site)p(.)17f(.) 767.21 256.75 0.000 9 18 (site)p(.)18f(landscape) 767.38 256.92 0.000 10 19 (site)p(.)19f(floristic richness) 768.70 258.24 0.000 10 20 (site)p(.)20f(cover) 768.75 258.29 0.000 10 21 (site)p(.)21f(t FST) 768.98 258.52 0.000 10 22 (site)p(.)22f(fires) 769.05 258.59 0.000 10 23 (site)p(.)23f(p FST) 769.34 258.88 0.000 10 24 (site)p(.)24f(H O) 769.36 258.91 0.000 10 25 (site)p(.)25f(age) 769.36 258.91 0.000 10 26 (site)p(.)26f(site) 775.60 265.15 0.000 15
Pereoglou, F., Population processes in an early successional heathland 110 Chapter 5: Population dynamics
this study. The low level of movement between populations suggests that recruitment estimates primarily represent recruitment to the captured age class resulting from locally born individuals. The demographic analyses revealed information about genetic processes that otherwise would have remained unknown or supposition. The genetic site variables are responsive to recruitment patterns: for example, spatial and temporal genetic differentiation reflects the role of recruitment in genetic drift.
5.5.1 Internal population processes: limited immigration, neutral sex effect, age-structured survival
Recruitment by reproduction has been found to be more important than immigration in other disturbed habitats (Pocock et al., 2004, Dooley and Bowers, 1998). Reproduction as major source of recruitment reduces the need for migrants to successfully track suitable habitat within their dispersal distance and is one mechanism by which population recovery can proceed after perturbation (Lindenmayer et al., 2005, Banks et al., 2011a). We propose two plausible models of population dynamics for early successional specialists – Model I ‘colonization model’ where individuals persist in recently burned habitat and move large distances to newly suitable habitat (recruitment due to immigration); and Model II ‘persistence model’ where species are present across the landscape at very low densities but irruptions occur due to increases in reproduction and survival after fire (recruitment due to locally born individuals). Our results do not resolve these processes, however they show that within a few years of recent fire, populations are discrete and exchange few migrants. While populations appear to be isolated during the period of study (i.e. no movement between sampled populations, limited immigration, high population- specific FST), previous work on this species demonstrates increased genetic connectivity in association with recent fire (Pereoglou et al., 2013). One possibility is that fire promotes movement and colonization of newly suitable burnt habitat, but once established, populations then become relatively isolated until the next fire. We found populations with the highest rates of recruitment were more genetically diverse and were less differentiated from other populations. When recruitment is due to migration, alleles added to the population reduce genetic differentiation between populations. However, our result of spatial genetic structure despite limited immigration suggests populations with higher recruitment potentially experience less genetic drift. Therefore, changes in animal movement over time may be a necessary strategy of inhabiting ephemeral patches of early successional habitat (e.g. Model I followed by Model II as succession proceeds) and genetic diversity (and differentiation between years within populations) is a consequence of these recruitment and survival patterns.
Survival is directly affected by dispersal (Driscoll et al., 2014), and rates can be specific to age classes, sexes, or both. When dispersal is not biased towards either sex, the processes
Pereoglou, F., Population processes in an early successional heathland 111 Chapter 5: Population dynamics underpinning survival operate equally on both male and female individuals. For species dependent on early successional habitats, an absence of sex-biased dispersal is an advantageous strategy because greater colonization ability of one sex compared to the other would increase the chance of colonizing newly suitable unoccupied habitat without any potential mates (Pereoglou et al., 2013). As a consequence, sex-biased dispersal is not favored under high demographic stochasticity (Gros et al., 2008) which can be associated with an unpredictable environment. Furthermore, an age-structured population where offspring disperse is beneficial for adults to maximize reproductive potential by retaining access to (temporally) limited resources (e.g. food, refugia, or mates). If immature individuals die or disperse before breeding, then reproductive output is mainly due to resident adults in the source populations (Paradis et al., 1993).
5.5.2 External population processes: fire and landscape structure In our modelling, site-level habitat quality and landscape-level processes (i.e. disturbance) were more important influences on survival than landscape structure (habitat area and heterogeneity). This would be expected under a Model II scenario where post-fire increases in reproduction and survival are driven by site-level processes including the distribution of critical resources. We found an increased incidence of fire and habitat features associated with fire affected survival in (mainly) predictable ways. Reduced cover and high floristic diversity was associated with a higher survival rate. Cover and floristic richness are indicators of local fire history because both change in expected ways following disturbance; cover increases and floristic richness decreases with vegetation succession in heathland. At a certain threshold, cover and floristic richness can become critical resources. What constitutes a critical resource is dependent upon a species’ life history and habitat requirements (Morrison et al., 2006). The influence of fire on survival has been linked to the effects of fire on critical resources across multiple taxa. When fire increases the critical resource, increases in survival also occur (Rota et al., 2014). When the critical resource is unaffected by fire, or becomes sub-optimal relative to pre-fire conditions, survival rates remain constant (Banks et al., 2011b, D'Souza et al., 2013, McDonough and Rexstad, 2005). When fire decreases the critical resource, a negative impact of fire on survival has observed (Pardon et al., 2003, Pons et al., 2003). The negative relationship between survival and density of dead shrubs was surprising despite their function as a critical shelter resource. In heathland a collapsed dead shrub layer is created and maintained by the frequency of fire events (Williams, 1995, Specht, 1981). Reduced survival in populations with a high density of dead shrubs may occur through increased intraspecific competition for shelter sites, or increased mortality through predation by snakes. Habitat quality also can be measured in terms of landscape structure. At this post-fire period of study, landscape effects on survival and recruitment were not important; models
Pereoglou, F., Population processes in an early successional heathland 112 Chapter 5: Population dynamics including landscape structure were relatively unsupported (Table 5.3 and 5.4). In the context of Model I and II, landscape effects are potentially critical during the colonisation phase, but not during the 'persistence' phase. For example, if individuals were required to make large-scale movements to colonise suitable habitat and immigration were a significant source of recruits (i.e. Model I), landscape structure should be a strong influence on recruitment because animals would need to traverse unsuitable habitat and colonization would possibly be dependent on patch size (i.e. less chance of successfully colonizing small patches either because they are more difficult to find or because they are already occupied).
5.5.3 Conclusion
We tested the hypothesis that species with genetically structured populations were more likely to have demography influenced by site-level processes than landscape-level processes, because populations that were discrete were likely to be strongly influenced by local resources. Populations appeared isolated, with demographic differences driven by effects of fire history on survival, most likely through impacts on habitat quality. Our modelling suggests that in early successional communities, (i) reproduction is a major source of recruitment and not immigration, (ii) variation in recruitment rate is due to spatial genetic structure and genetic composition of individuals in populations, and (iii) survival rates are strongly influenced by spatial variation in habitat quality as a result of previous fires and to a lesser extent landscape structure. These findings indicate fire provides an important ecological condition for early successional species, and that the importance of fire may be much broader than commonly appreciated for sustaining demographic processes.
5.6 Acknowledgement
We most gratefully acknowledge the Wreck Bay Aboriginal Community and the Wandandian People of the Dharawal and Dhurga language groups on whose lands this study was conducted. We thank Wendy Hartman, Alfred Wellington, Tony Carter, Geoff Kay, Steve Phillips, Marama Hopkins, Nick Dexter, Karen Viggers, Sally Townley, Les Mitchell and Martin Westgate. We also thank Axios Review for referee comments that greatly improved this manuscript. This research was funded by the Norman Wettenhall Foundation, MA Ingram Trust, Australian Academy of Science, WV Scott Charitable Trust.
5.7 References
Aars, J. & Ims, R. A. (2002) Intrinsic and climatic determinants of population demography: the winter dynamics of tundra voles. Ecology, 83, 3449-3456.
Pereoglou, F., Population processes in an early successional heathland 113 Chapter 5: Population dynamics
Abadi, F., Gimenez, O., Ullrich, B., Arlettaz, R. & Schaub, M. (2010) Estimation of immigration rate using integrated population models. Journal of Applied Ecology, 47, 393-400. Banks, S. C., Cary, G. J., Smith, A. L., Driscoll, D. A., Lindenmayer, D. B., Davies, I. D., Gill, A. M. & Peakall, R. (2013) How does ecological disturbance influence genetic diversity? Trends in Ecology & Evolution, 28, 670-679. Banks, S. C., Dujardin, M., McBurney, L., Blair, D., Barker, M. & Lindenmayer, D. B. (2011a) Starting points for small mammal population recovery after wildfire: recolonisation or residual populations? Oikos, 120, 26-37.
Banks, S. C., Knight, E. J., McBurney, L., Blair, D. & Lindenmayer, D. B. (2011b) The effects of wildfire on mortality and resources for an arboreal marsupial: resilience to fire events but susceptibility to fire regime change. PLoS ONE, 6, 1-12. Barbour, M. S. & Litvaitis, J. A. (1993) Niche dimensions of New England cottontails in relation to habitat patch size. Oecologia, 95, 321-327. Betts, M. G., Hagar, J. C., Rivers, J. W., Alexander, J. D., McGarigal, K. & McComb, B. C. (2010) Thresholds in forest bird occurrence as a function of the amount of early-seral broadleaf forest at landscape scales. Ecological Applications, 20, 2116-2130. Bowers, M. A., Gregario, K., Brame, C. J., Matter, S. F. & Dooley, J. L., Jr. (1996) Use of space and habitats by meadow voles at the home range, patch and landscape scales. Oecologia, 105, 107-115. Boyce, M. S., Haridas, C. V., Lee, C. T. & NCEASStochasticDemographyWorkingGroup. (2006) Demography in an increasingly variable world. Trends in Ecology & Evolution, 21, 141-148. Breed, B. & Ford, F. (2007) Native Mice and Rats, CSIRO Publishing, Collingwood. Burnham, K. P. & Anderson, D. R. (2002) Model selection and multi-modal inference: a practical information-theoretic approach., Springer, New York. Choquet, R., Lebreton, J. D., Gimenez, O., Reboulet, A. M. & Pradel, R. (2009) U‐ CARE: Utilities for performing goodness of fit tests and manipulating CApture–REcapture data. Ecography, 32, 1071 - 1074. Cooch, E. & White, G. (2014) Program MARK: A gentle introduction. Available online in PDF format at: http://www.phidot.org/software/mark/docs/book/. D'Souza, J. B., Whittington, A., Dickman, C. R. & Leung, L. K. P. (2013) Perfect storm: demographic responses of an irruptive desert mammal to prescribed burns following flooding rain. Austral Ecology, 38, 765-776.
Pereoglou, F., Population processes in an early successional heathland 114 Chapter 5: Population dynamics
Donner, D. M., Ribic, C. A. & Probst, J. R. (2010) Patch dynamics and the timing of colonisation-abandonment events by male Kirtland's warblers in an early succession habitat. Biological Conservation, 143, 1159-1167. Dooley, J. L., Jr. & Bowers, M. A. (1998) Demographic responses to habitat fragmentation: experimental tests at the landscape and patch scale. Ecology, 79, 969-980. Driscoll, D. A., Banks, S. C., Barton, P. S., Ikin, K., Lentini, P., Lindenmayer, D. B., Smith, A. L., Berry, L. E., Burns, E. L., Edworthy, A., Evans, M. J., Gibson, R., Heinsohn, R., Howland, B., Kay, G., Munro, N., Scheele, B. C., Stirnemann, I., Stojanovic, D., Sweaney, N., Villasenor, N. R. & Westgate, M. (2014) The trajectory of dispersal research in conservation biology. Systematic review. PLoS ONE, 9, e95053. ESRI (2009) ArcMap 9.2. Environmental Systems Resource Institute, California. Foll, M. & Gaggiotti, O. (2006) Identifying the environmental factors that determine the genetic structure of populations. Genetics, 174, 875-891. Fox, B. J. (1982) Fire and mammalian secondary succession in an Australian coastal heath. Ecology, 63, 1332-1341. Fox, B. J. (2008) Eastern chestnut mouse, Pseudomys gracilicaudatus. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 634-635. Reed New Holland, Sydney, Australia.
Fox, B. J. & Kemper, C. M. (1982) Growth and development of Pseudomys gracilicaudatus (Rodentia: Muridae) in the laboratory. Australian Journal of Zoology, 30, 175-185. Goudet, J. (2001) FSTAT, a program to estimate and test gene diversities and fixation indices (version 2.9.3). Available from http://www.unil.ch/izea/softwares/fstat.html. Gros, A., Hovestadt, T. & Poethke, H. J. (2008) Evolution of sex-biased dispersal: The role of sex-specific dispersal costs, demographic stochasticity, and inbreeding Ecological Modelling, 219, 226-233. Hanski, I. (2009) Metapopulations and spatial population processes. The Princeton guide to ecology (ed S. A. Levin), pp. 177-185. Princeton University Press, New Jersey.
Holland, G. J. & Bennett, A. F. (2010) Habitat fragmentation disrupts the demography of a widespread native mammal. Ecography, 33, 841-853. Hutto, R. L. (1995) Composition of bird communities following stand-replacement fires in northern Rocky Mountain (USA) conifer forests. Conservation Biology, 9, 1041-1058. Ives, A. R. (2009) Density dependence and single species population dynamics. The Princeton guide to ecology (ed S. A. Levin), pp. 166-171. Princeton University Press, New Jersey. Lebreton, J.-D., Burnham, K. P., Clobert, J. & Anderson, D. R. (1992) Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs, 62, 67-118.
Pereoglou, F., Population processes in an early successional heathland 115 Chapter 5: Population dynamics
Lima, M., Julliard, R., Stenseth, N. & Jaksic, F. M. (2001) Demographic dynamics of a neotropical small rodent (Phyllotis darwini): feedback structure, predation and climatic factors. Journal of Animal Ecology, 70, 761-775. Lindenmayer, D. B., Cunningham, R. B. & Peakall, R. (2005) The recovery of populations of bush rat Rattus fuscipes in forest fragments following major population reduction. Journal of Applied Ecology, 42, 649-658. McDonough, T. J. & Rexstad, E. (2005) Short-term demographic response of the red-backed vole to spruce beetle infestations in Alaska. Journal of Wildlife Management, 69, 246- 254.
Mills, L. S. (2013) Conservation of wildlife populations: demography, genetics and management. 2nd Edn, John Wiley and Sons, Ltd. Morrison, M. L., Marcot, B. G. & Mannan, R. W. (2006) Wildlife-habitat relationships: concepts and applications, Island Press, Washington. Nupp, T. E. & Swihart, R. K. (1996) Effect of forest patch area on population attributes of white-footed mice (Peromyscus leucopus) in fragmented landscapes. Canadian Journal of Zoology-Revue Canadienne De Zoologie, 74, 467-472. Paetkau, D., Slade, R., Burden, M. & Estoup, A. (2004) Direct real-time estimation of migration rate using assignment methods: a simulation-based exploration of accuracy and power. Molecular Ecology, 13, 55-65. Paradis, E., Guedon, G. & Pradel, R. (1993) Estimation of sex- and age-related survival rates in a microtine population. Journal of Wildlife Management, 57, 158-163. Pardon, L. G., Brook, B. W., Griffiths, A. D. & Braithwaite, R. W. (2003) Determinants of survival for the northern brown bandicoot under a landscape-scale fire experiment. Journal of Animal Ecology, 72, 106-115. Park, S. D. E. (2001) Trypanotolerance in West African Cattle and the Population Genetic Effects of Selection. University of Dublin, Dublin, Ireland. Peakall, R. & Smouse, P. E. (2006) GenAlEx 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes, 6, 288-295. Pereoglou, F., Lindenmayer, D. B., MacGregor, C., Ford, F., Wood, J. & Banks, S. C. (2013) Landscape genetics of an early successional specialist in a disturbance-prone environment. Molecular Ecology, 22, 1267-1281. Pereoglou, F., MacGregor, C., Banks, S. C., Ford, F. D., Wood, J. T. & Lindenmayer, D. B. (2011) Refuge site selection by the eastern chestnut mouse in recently burned heath. Wildlife Research, 38, 290-298.
Pereoglou, F., Population processes in an early successional heathland 116 Chapter 5: Population dynamics
Pierson, J. C., Allendorf, F. W., Drapeau, P. & Schwartz, M. K. (2013) Breed locally, disperse globally: fine-scale genetic structure despite landscape-scale panmixia in a fire- specialist. PLoS ONE, 8, e67248. Piry, S., Alapetite, A., Cornuet, J. M., Paetkau, D., Baudouin, L. & Estoup, A. (2004) GeneClass2: A software for genetic assignment and first-generation migrant detection. Journal of Heredity, 95, 536-539. Pocock, M. J. O., Searle, J. B. & White, P. C. L. (2004) Adaptations of animals to commensal habitats: population dynamics of house mice Mus musculus domesticus on farms. Journal of Animal Ecology, 73, 878-888.
Pons, P., Henry, P.-Y., Gargallo, G., Prodon, R. & Lebreton, J.-D. (2003) Local survival after fire in Mediterranean shrublands: combining capture-recapture data over several bird species. Population Ecology, 45, 187-196. Pradel, R. (1996a) Utilization of capture-mark-recapture for the study of recruitment and population growth rate. Biometrics, 52, 703-709. Pradel, R. (1996b) Utilization of capture-mark-recapture for the study of recruitment and population growth rate. Biometrics, 52, 703-709. Pryde, M. A., O'Donnell, C. F. J. & Barker, R. J. (2005) Factors influencing survival and long- term population viability of New Zealand long-tailed bats (Chalinolobus tuberculatus): implications for conservation. Biological Conservation, 126, 175-185. Rannala, B. & Mountain, J. L. (1997) Detecting immigration using multilocus genotypes. Proceedings of the National Academy of Sciences, 94, 9197-9221. Rota, C. T., Millspaugh, J. J., Rumble, M. A., Lehman, C. P. & Kesler, D. C. (2014) The role of wildfire, prescribed fire, and mountain pine beetle infestations on the population dynamics of black-backed woodpeckers in the Black Hills, South Dakota. PLoS ONE, 9, e94700. Sandercock, B. K. (2006) Estimate of demographic parameters from live-encounter data: a summary review. Journal of Wildlife Management, 70, 1504-1520.
Schaub, M., Jakober, H. & Stauber, W. (2013) Strong contribution of immigration to local population regulation: evidence from a migratory passerine. Ecology, 94, 1828-1838. Schaub, M., Reichlin, T. S., Abadi, F., Kery, M., Jenni, L. & Arlettaz, R. (2012) The demographic drivers of local population dynamics in two rare migratory birds. Oecologia, 168, 97-108. Sibly, R. M., Collett, D., Promislow, D. E. L., Peacock, D. J. & Harvey, P. H. (1997) Mortality rates of mammals. Journal of Zoology, London, 243, 1-12. Soga, M. & Koike, S. (2013) Large forest patches promote breeding success of a terrestrial mammal in urban landscapes. PLoS ONE, 8, e51802.
Pereoglou, F., Population processes in an early successional heathland 117 Chapter 5: Population dynamics
Specht, R. L. (1981) Responses to fire in heathlands and related shrublands. Fire and the Australian Biota (eds A. M. Gill, R. H. Groves & I. R. Noble), pp. 395-415. Australian Academy of Science, Canberra. Taws, N. (1997) Vegetation survey and mapping of Jervis Bay Territory. Report to Environment Australia. Canberra. Templeton, A. R., Brazeal, H. & Neuwald, J. L. (2011) The transition from isolated patches to a metapopulation in the eastern collared lizard in response to prescribed fires. Ecology, 92, 1736-1747. Templeton, A. R., Robertson, R. J., Brisson, J. & Strasburg, J. (2001) Disrupting evolutionary processes: the effect of habitat fragmentation on collared lizards in the Missouri Ozarks. Proceedings of the National Academy of Sciences, 98, 5426-5432. Troyer, E. M., Cameron Devitt, S. E., Sunquist, M. E., Goswami, V. R. & Oli, M. K. (2014) Density dependence or climatic variation? Factors influencing survival, recruitment, and population growth rate of Virginia opossums. Journal of Mammalogy, 95, 421-430. Umapathy, G. & Kumar, A. (2000) The demography of the lion-tailed macaque (Macaca silenus) in rain forest fragments in the Anamalai Hills, South India. Primates, 41, 119- 126. Villafuerte, R., Litvaitis, J. A. & Smith, D. F. (1997) Physiological responses by lagomorphs to resource limitations imposed by habitat fragmentation: implications for condition- sensitive predation. Canadian Journal of Zoology-Revue Canadienne De Zoologie, 75, 148-151. Villella, R. F., Smith, D. R. & Lemarie, D. P. (2004) Estimating survival and recruitment in a freshwater mussel population using mark-recapture technique. The American Midland Naturalist, 151, 114-133. White, G. C. & Burnham, K. P. ( 1999) Program MARK: Survival estimation from populations of marked animals. Bird Study, 46 Supplement, 120-138. Williams, D. G. (1995) Heaths and scrubs. Jervis Bay: A Place of Cultural, Scientific and Educational Value. Kowari 5 (eds G. Cho, A. Georges, R. Stoutjesdijk & R. Longmore), pp. 83-90. Australian Nature Conservation Agency, Canberra.
Pereoglou, F., Population processes in an early successional heathland 118
Chapter 6: Conclusion
Conclusion
Pereoglou, F., Population processes in an early successional heathland 119 Chapter 6: Conclusion
6.1 Population processes of the eastern chestnut mouse
This thesis is an investigation of the animal population processes occurring in an early successional heathland. The eastern chestnut mouse was used as a case study for this research. For an early successional specialist such as the eastern chestnut mouse, the opportunity to compare processes operating in populations before and/or after disturbance is limited by the unpredictable occurrence of wildfires, the species’ occurrence at low density, and a ‘patchy’ distribution across its geographic range. We know this species responds positively to fire, but we don’t understand the underlying demographic processes involved, nor do we understand the contribution of individual movements, disturbance history, and habitat quality. The research presented here provides new knowledge on the population ecology of the eastern chestnut mouse. This project, for the first time, elucidated mechanisms of population persistence in an early successional heathland through addressing a series of research questions aimed at identifying the resources used for refuge and associated behaviours, predictors of species occurrence, spatial patterns of genetic structure, and drivers of population dynamics in spatially and temporally variable environment. I provide a summary of the previous four research chapters and discuss the mechanisms of population persistence in early successional communities, including the specific role of fire. Chapter 2 described habitat features and structures selected by the eastern chestnut mouse as refuge sites, and patterns of refuge site fidelity in recently burned heathland. Refuge sites consisted of taller, denser vegetation than randomly selected control sites, and were associated with specific structures including collapsed dead shrubs, Xanthorrhea and Gymnoschoenus species. There were no differences in the sites selected by males and females. However, there was evidence for a primary–secondary model of refuge use. Primary refuge sites contained a nesting structure (i.e. pad or spherical nest of woven, dried grass) and were the densest sites selected. Seasonal variation in habitat selection and fidelity to sites was attributed to reproductive constraints, intra-specific competition, and environmental conditions. Mice used multiple, rarely occupied refuges, and were mainly solitary although they infrequently ‘shared’ sites without cohabitation. The major finding in Chapter 2 was that refuge sites were highly flammable and would not provide protection from fire. Mice did not select fire-resistant refugia (i.e. rock crevasses or tree roots), or use behaviours associated with fire avoidance (i.e. burrowing). These resource requirements and patterns of use have clear implications for fire management and prescribed burning practices (see ‘The role of fire’ below). Chapter 3 evaluated the effects of disturbance, landscape structure, and site-level habitat quality on eastern chestnut mouse site occupancy in an early successional ecosystem. This chapter introduced the issue of scale and assessed spatial patterns of landscape suitability and site-level habitat requirements affecting early successional fauna. Landscape configuration was a strong influence on site occupancy; occupancy increased with habitat patch area and decreased
Pereoglou, F., Population processes in an early successional heathland 120 Chapter 6: Conclusion with landscape heterogeneity. Resource availability at a local (site-level) scale also was significant. The results affirmed the importance of a collapsed dead shrub layer and indicated the number of dead shrubs increased with burning in the last major wildfire. Although rock crevasses were not identified as refuge sites, rock cover was positively correlated with site occupancy. No predictable patterns of small mammal species co-occurrence were found, contrary to the mammalian secondary succession model. No direct effect of fire was detected. The major finding in Chapter 3 was that fire affected population processes indirectly and site occurrence was a multi-scale course of action. These results have implications for land management including the extent of prescribed fire, clearing or habitat modification, and surrounding land use, and also for experimental design as a guide for survey work targeting populations of the eastern chestnut mouse. Chapter 4 assessed the effects of landscape variation and recent disturbance history on dispersal by the eastern chestnut mouse. Significant genetic structure was identified, and positive spatial autocorrelation between genotypes was detected up to 2–3 km, indicating dispersal was restricted relative to the overall spatial scale of the study. As predicted, there was no strong evidence for female philopatry and male dispersal. The influence of different habitat- and fire-specific landscape resistance scenarios on genetic connectivity was tested. Recently burned vegetation had greater conductance for gene flow than unburnt habitat, but variation in habitat quality between vegetation types and occupied patches had no effect on gene flow. This was attributed to fire improving habitat suitability for early successional species, and habitat suitability improving genetic connectivity across the landscape. An alternative was that dispersal in burnt areas increased in association with post-fire colonization before returning to levels described under a ‘null landscape’ model. The major finding of Chapter 4 was that fire assisted in structuring populations of an early successional species. Dispersal through unsuitable habitat was possible and the effects of landscape heterogeneity were likely dependent upon the extent of dissimilarity was from a more ‘natural’ state. Chapter 5 investigated two major components of population growth in the eastern chestnut mouse – survival and recruitment, and how these key demographic parameters vary across the landscape in response to fire history related environmental factors. In this section, demographic and genetic analyses were used to quantify survival, recruitment and immigration rates. Survival varied across sites, with supported models including disturbance history and resources indicative of local fire history (collapsed dead shrubs, cover, floristic richness). Compared to local reproduction, immigration contributed very little to recruitment, suggesting populations are demographically discrete at this post-fire stage. Sites with high recruitment also had higher genetic diversity and lower temporal genetic structure, consistent with reduced genetic drift. The key findings of this chapter suggests a two-phase model of spatial population dynamics, where post-fire colonization is followed by locally driven population dynamics, in which survival rates are influenced by fire-history effects on habitat quality. These results have implications for land
Pereoglou, F., Population processes in an early successional heathland 121 Chapter 6: Conclusion management including the application of fire to maintain critical resources associated with increased survival, and promote genetic connectivity between local populations of species inhabiting post-fire early successional communities by initiating immigration that is otherwise relatively limited.
6.2 Mechanisms of population persistence in an early successional community The mechanisms used by animals to persist in an early successional ecosystem are discussed in this section. I propose two possible persistence scenarios for early successional species: either (I) individuals are present across the landscape at very low densities but increases in density follow increases in reproduction and survival post-fire (‘persistence model’), or (II) individuals are present in recently burned habitats and move large distances to newly suitable habitat when it becomes available post-fire (‘colonisation model’). The research I have presented in this thesis provides evidence for both potential scenarios. The maintenance of multiple, rarely occupied refuge sites that maximises access to resources that are ‘patchy’ or dispersed, and potentially limited over time is a mechanism underlying the colonisation model of population persistence because refuge sites are likely to be burned by fire, limited by successional changes in vegetation over time, and therefore temporary over longer time scales. The use of existing refuge sites by multiple animals (i.e. inheritance and/or temporally segregated use of refuge sites) to avoid associated costs of seeking suitable new sites in spatially and temporally variable habitat is indicative of the persistence model where population density is (probably) positively associated with the availability of suitable, unoccupied shelter resources. Landscape structure (patch area and landscape heterogeneity) has the potential to mitigate the effects of disturbance for early successional species. A generalist dispersal strategy where animal movement is not sex-biased, or restricted to particular vegetation types or landscape attributes is an indicator of a colonisation model of population persistence because it is the disperser’s ability to follow fire determines whether newly suitable habitat is located. Local reproduction by residents would be expected under a persistence model as an advantageous strategy for increasing population density compared to immigration when the need for migrants to track suitable (available) habitat within dispersal distance in post-fire landscapes can be avoided. In this thesis, I have studied population processes that occur over multiple scales. For example, animal movement, including dispersal and migration, occurs over a landscape scale, whereas equally important mechanisms involving habitat selection, colonization and recruitment occur at a local (site-level) scale. The obvious limitation to the research presented here is that the effects of time and temporal successional changes on refuge site selection and use, predictors of species occurrence, spatial patterns of genetic structure, and drivers of population dynamics currently remain unknown.
Pereoglou, F., Population processes in an early successional heathland 122 Chapter 6: Conclusion 6.2.1 The role of fire Changes to natural fire regimes associated with climate change are expected to result in more frequent disturbance events. Our understanding of the positive aspects of fire-related disturbance is rudimentary and must improve if we seek to conserve biodiversity in early successional environments. Fire can have an indirect effect on population processes in early successional communities. As discussed throughout this thesis, habitat elements created and/or maintained by fire (i.e. a collapsed dead shrub layer, cover, floristic richness, etc.) influence the selection of diurnal refugia, success of site occupancy, and demographic rates. In general, when significant, the direct effect of fire was positive; i.e. fire promotes genetic connectivity between local populations of early successional fauna, and possibly immigration although this later process is unconfirmed. Future research in this subject area might include quantifying the effect of fire regime and landscape structure on the critical resources identified in this thesis, demographic rates, and dispersal strategy of the eastern chestnut mouse.
Pereoglou, F., Population processes in an early successional heathland 123
Consolidated references
Consolidated references
Pereoglou, F., Population processes in an early successional heathland 124 Chapter 7: Consolidated references Consolidated References
Aars, J. & Ims, R. A. (2002) Intrinsic and climatic determinants of population demography: the winter dynamics of tundra voles. Ecology, 83, 3449-3456. Abadi, F., Gimenez, O., Ullrich, B., Arlettaz, R. & Schaub, M. (2010) Estimation of immigration rate using integrated population models. Journal of Applied Ecology, 47, 393-400. Baguette, M. & Van Dyck, H. (2007) Landscape connectivity and animal behavior: functional grain as a key determinant for dispersal. Landscape Ecology, 22, 1117-1129. Baker, J., Whelan, R. J., Evans, L., Moore, S. & Norton, M. (2010) Managing the ground parrot in its fiery habitat in south-eastern Australia. Emu, 110, 279-284. Banks, S. C., Cary, G. J., Smith, A. L., Driscoll, D. A., Lindenmayer, D. B., Davies, I. D., Gill, A. M. & Peakall, R. (2013) How does ecological disturbance influence genetic diversity? Trends in Ecology & Evolution, 28, 670-679. Banks, S. C., Dujardin, M., McBurney, L., Blair, D., Barker, M. & Lindenmayer, D. B. (2011a) Starting points for small mammal population recovery after wildfire: recolonisation or residual populations? Oikos, 120, 26-37. Banks, S. C., Finlayson, G. R., Lawson, S. J., Lindenmayer, D. B., Paetkau, D., Ward, S. J. & Taylor, A. C. (2005a) The effects of habitat fragmentation due to forestry plantation establishment on the demography and genetic variation of a marsupial carnivore, Antechinus agilis. Biological Conservation, 122, 581-597. Banks, S. C., Knight, E. J., McBurney, L., Blair, D. & Lindenmayer, D. B. (2011b) The effects of wildfire on mortality and resources for an arboreal marsupial: resilience to fire events but susceptibility to fire regime change. PLoS ONE, 6, 1-12. Banks, S. C., Lindenmayer, D. B., Ward, S. J. & Taylor, A. C. (2005b) The effects of habitat fragmentation via forestry plantation establishment on spatial genotypic structure in the small marsupial carnivore, Antechinus agilis. Molecular Ecology, 14, 1667-1680. Banks, S. C. & Peakall, R. (2012) Genetic spatial autocorrelation can readily detect sex-biased dispersal. Molecular Ecology, 14, 2092-2105. Banks, S. C., Ward, S. J., Lindenmayer, D. B., Finlayson, G. R., Lawson, S. J. & Taylor, A. C. (2005c) The effects of habitat fragmentation on the social kin structure and mating system of the agile antechinus, Antechinus agilis. Molecular Ecology, 14, 1789-1801. Barbour, M. S. & Litvaitis, J. A. (1993) Niche dimensions of New England cottontails in relation to habitat patch size. Oecologia, 95, 321-327. Baur, A. & Baur, B. (1992) Effect of corridor width on animal dispersal: a simulation study. Global Ecology and Biogeography Letters, 2, 52-56. Betts, M. G., Hagar, J. C., Rivers, J. W., Alexander, J. D., McGarigal, K. & McComb, B. C. (2010) Thresholds in forest bird occurrence as a function of the amount of early-seral broadleaf forest at landscape scales. Ecological Applications, 20, 2116-2130. Pereoglou, F., Population processes in an early successional heathland 125 Chapter 7: Consolidated references
Bond, W. J. & Keeley, J. E. (2005) Fire as a global herbivore: the ecology and evolution of flammable ecosystems. Trends in Ecology & Evolution, 20, 387-394. Bowers, M. A., Gregario, K., Brame, C. J., Matter, S. F. & Dooley, J. L., Jr. (1996) Use of space and habitats by meadow voles at the home range, patch and landscape scales. Oecologia, 105, 107-115. Bowler, D. E. & Benton, T. G. (2005) Causes and consequences of animal dispersal strategies: relating individual behaviour to spatial dynamics. Biological Reviews, 80, 205-225. Boyce, M. S., Haridas, C. V., Lee, C. T. & NCEASStochasticDemographyWorkingGroup. (2006) Demography in an increasingly variable world. Trends in Ecology & Evolution, 21, 141-148. Bradstock, R. A., Williams, J. & Gill, A. M. (2012) Flammable Australia: The Fire Regimes and Biodiversity of a Continent. Cambridge University Press, Cambridge. Braithwaite, R. W. (1982) Small marginal groups of Rattus lutreolus in good and poor habitat. Journal of Zoology, 198, 529-532. Braithwaite, R. W. (1990) Australia's unique biota: implications for ecological processes. Journal of Biogeography, 17, 347-354. Breed, B. & Ford, F. (2007) Native Mice and Rats, CSIRO Publishing, Collingwood. Bruford, M. W., Hanotte, O., Brookfield, J. K. Y. & Burke, T. (1988) Multilocus and single- locus DNA fingerprinting. Molecular Genetic Analysis of Populations: A Practical Approach (ed A. R. Hoelzel), pp. 287-336. IRL Press, Oxford. Bubela, T. M. & Happold, D. C. D. (1993) The social organization and mating system of an Australian sub-alpine rodent, the broad-toothed rat Mastacomys fuscus Thomas. Wildlife Research, 20, 405-417. Burel, F. & Baudry, J. (2003) Landscape ecology: concepts, methods and applications, Science Publishers, Enfield, USA. Burnham, K. P. & Anderson, D. R. (2002) Model selection and multi-modal inference: a practical information-theoretic approach., Springer, New York. Catling, P. C. (1986) Rattus lutreolus, colonizer of heathland after fire in the absence of Pseudomys species. Australian Wildlife Research, 13, 127-139. Catling, P. C. (1991) Ecological effects of prescribed burning practices on the mammals of southeastern Australia. Conservation of Australia's Forest Fauna (ed D. Lunney), pp. 353-363. Royal Zoological Society of NSW, NSW. Catling, P. C. & Newsome, A. E. (1981) Responses of the Australian vertebrate fauna to fire: an evolutionary approach. Fire and the Australian Biota (eds A. M. Gill, R. H. Groves & I. R. Noble), pp. 273 - 310. Australian Academy of Science, Canberra. Chesson, P. (2000) Mechanisms of maintenance of species diversity. Annual Review of Ecology and Systematics, 31, 343-366.
Pereoglou, F., Population processes in an early successional heathland 126 Chapter 7: Consolidated references Choquet, R., Lebreton, J. D., Gimenez, O., Reboulet, A. M. & Pradel, R. (2009) U‐ CARE: Utilities for performing goodness of fit tests and manipulating CApture–REcapture data. Ecography, 32, 1071 - 1074. Clarke, M. F. (2008) Catering for the needs of fauna in fire management: science or just wishful thinking? Wildlife Research, 35, 385-394. Cockburn, A. (1978) Distribution of Pseudomys shortridgei (Muridae: Rodentia) and its relevance to that of other heathland Pseudomys. Australian Wildlife Research, 5, 213- 219. Cockburn, A. (1981a) Diet and habitat preference of the silky desert mouse Pseudomys apodemoides (Rodentia). Australian Wildlife Research, 8, 475-97. Cockburn, A. (1981b) Population processes of the silky desert mouse Pseudomys apodemoides (Rodentia) in mature heathlands. Australian Wildlife Research, 8, 499-514. Cockburn, A., Braithwaite, R. W. & Lee, A. K. (1981) The response of the heath rat, Pseudomys shortridgei, to pyric succession - a temporally dynamic life-history strategy. Journal of Animal Ecology, 50, 649-666. Cockburn, A. & Lazenby-Cohen, K. A. (1992) Use of nest trees by Antechinus stuartii, a semelparous lekking marsupial. Journal of Zoology, London, 226, 657-680. Cockburn, A., Scott, M. P. & Scotts, D. J. (1985) Inbreeding avoidance and male-biased natal dispersal in Antechinus spp. (Marsupialia: Dasyuridae). Animal Behaviour, 33. Collett, D. (2002) Modelling Binary Data, Chapman and Hall, London. Connell, J. H. & Slatyer, R. O. (1977) Mechanisms of succession in natural communities and their role in community stability and organization. American Naturalist, 111, 1119- 1144. Conover, W. J. (1999) Practical Nonparametric Statistics, 3rd Edition, Wiley, New York. Cooch, E. & White, G. (2014) Program MARK: A gentle introduction. Available online in PDF format at: http://www.phidot.org/software/mark/docs/book/. Crowther, M. S. (2002) Distributions of species of the Antechinus stuartii-A. flavipes complex as predicted by bioclimatic modelling. Australian Journal of Zoology, 50, 77-91. Crowther, M. S. & Braithwaite, R. W. (2008) Brown antechinus, Antechinus stuartii. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 94-96. Reed New Holland, Sydney, Australia. Cushman, S. A., McKelvey, K. S., Hayden, J. & Schwartz, M. K. (2006) Gene flow in complex landscapes: Testing multiple hypotheses with causal modeling. American Naturalist, 168, 486-499. D'Souza, J. B., Whittington, A., Dickman, C. R. & Leung, L. K. P. (2013) Perfect storm: demographic responses of an irruptive desert mammal to prescribed burns following flooding rain. Austral Ecology, 38, 765-776.
Pereoglou, F., Population processes in an early successional heathland 127 Chapter 7: Consolidated references
Delmoral, R. & Bliss, L. C. (1993) Mechanisms of primary succession: insights resulting from the eruption of Mount St Helens. Advances in Ecological Research, 24, 1-66. Di Stefano, J., Owen, L., Morris, R., Duff, T. & York, A. (2011) Fire, landscape change and models of small mammal habitat suitability at multiple spatial scales. Austral Ecology, 36, 638-649. Dickman, C. R. (2008) Agile antechinus, Antechinus agilis. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 83-84. Reed New Holland, Sydney, Australia. Dickman, C. R., Lunney, D. & Matthews, A. (2000) Ecological attributes and conservation of native rodents in New South Wales. Wildlife Research, 27, 347-355. Dickman, C. R., Predavec, M. & Downey, F. J. (1995) Long-range movements of small mammals in arid Australia: Implications for land management. Journal of Arid Environments, 31, 441-452. Donner, D. M., Ribic, C. A. & Probst, J. R. (2010) Patch dynamics and the timing of colonisation-abandonment events by male Kirtland's warblers in an early succession habitat. Biological Conservation, 143, 1159-1167. Dooley, J. L., Jr. & Bowers, M. A. (1998) Demographic responses to habitat fragmentation: experimental tests at the landscape and patch scale. Ecology, 79, 969-980. Driscoll, D. A., Banks, S. C., Barton, P. S., Ikin, K., Lentini, P., Lindenmayer, D. B., Smith, A. L., Berry, L. E., Burns, E. L., Edworthy, A., Evans, M. J., Gibson, R., Heinsohn, R., Howland, B., Kay, G., Munro, N., Scheele, B. C., Stirnemann, I., Stojanovic, D., Sweaney, N., Villasenor, N. R. & Westgate, M. (2014) The trajectory of dispersal research in conservation biology. Systematic review. PLoS ONE, 9, e95053. Driscoll, D. A. & Henderson, M. K. (2008) How many common reptile species are fire specialists? A replicated natural experiment highlights the predictive weakness of a fire succession model. Biological Conservation, 141, 460-471. Ebensperger, L. A., Sobrero, R., Campos, V. & Giannoni, S. M. (2008) Activity, range areas, and nesting patterns in the viscacha rat Octomys mimax. Journal of Arid Environments, 72, 1174-1183. ESRI (2009) ArcMap 9.2. Environmental Systems Resource Institute, California. Excoffier, L., Foll, M. & Petit, R. J. (2009) Genetic consequences of range expansions. Annual Review of Ecology, Evolution, and Systematics, 40, 481-501. Fahrig, L. (2007) Non-optimal animal movement in human-altered landscapes. Functional Ecology, 21, 1003-1015. Fahrig, L. & Paloheimo, J. (1988) Determinants of local population size in patchy habitats. Theoretical Population Biology, 34, 194-213. Flannigan, M. D., Krawchuk, M. A., de Groot, W. J., Wotton, B. M. & Gowman, L. M. (2009) Implications of changing climate for global wildland fire. International Journal of Wildland Fire, 18, 483-507.
Pereoglou, F., Population processes in an early successional heathland 128 Chapter 7: Consolidated references
Foll, M. & Gaggiotti, O. (2006) Identifying the environmental factors that determine the genetic structure of populations. Genetics, 174, 875-891. Ford, F. D. (2003) Conilurine rodent evolution: the role of ecology in modifying evolutionary consequences of environmental change. James Cook University, Townsville, Australia. Ford, F. D. (2008) Smokey mouse, Pseudomys fumeus. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 629-630. Reed New Holland, Sydney, Australia. Fox, B. J. (1982) Fire and mammalian secondary succession in an Australian coastal heath. Ecology, 63, 1332-1341. Fox, B. J. (1995) Eastern chestnut mouse Pseudomys gracilicaudatus. The Mammals of Australia (ed R. Strahan), pp. 601-3. Reed, Chatswood. Fox, B. J. (2008) Eastern chestnut mouse, Pseudomys gracilicaudatus. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 634-635. Reed New Holland, Sydney, Australia. Fox, B. J. & Archer, E. (1984) The diets of Sminthopsis murina and Antechinus stuartii (Marsupialia: Dasyuridae) in sympatry. Australian Wildlife Research, 11, 235-248. Fox, B. J. & Fox, M. D. (1981) A comparison of vegetation classifications as descriptors of small mammal habitat preference. Vegetation Classification in Australia: Proceedings of a Workshop Sponsored by CSIRO Division of Land Use Research, Canberra, October 1978 (eds A. N. Gillison & D. J. Anderson), pp. 166-180. CSIRO, Canberra. Fox, B. J. & Kemper, C. M. (1982) Growth and development of Pseudomys gracilicaudatus (Rodentia: Muridae) in the laboratory. Australian Journal of Zoology, 30, 175-185. Fox, B. J. & Monamy, V. (2007) A review of habitat selection by the swamp rat Rattus lutreolus (Rodentia: Muridae). Austral Ecology, 32, 837-849. Fox, B. J., Taylor, J. E. & Thompson, P. T. (2003) Experimental manipulation of habitat structure: a retrogression of the small mammal succession. Journal of Animal Ecology, 72, 927-940. Frank, P. A. & Layne, J. N. (1992) Nests and daytime refugia of cotton mice (Peromyscus gossypinus) and golden mice (Ochrotomys nuttalli) in south central Florida. American Midland Naturalist, 127, 21-30. Franklin, J. F., Lindenmayer, D., MacMahon, J. A., McKee, A., Magnuson, J., Perry, D. A., Waide, R. & Foster, D. (2000) Threads of continuity. Conservation in Practice, 1, 8-17. Friend, G. R. (1993) Impact of fire on small vertebrates in mallee woodlands and heathlands of temperate Australia: a review. Biological Conservation, 65, 99-114. Gardner-Santana, L. C., Norris, D. E., Fornadel, C. M., Hinson, E. R., Klein, S. L. & Glass, G. E. (2009) Commensal ecology, urban landscapes, and their influence on the genetic characteristics of city-dwelling Norway rats (Rattus norvegicus). Molecular Ecology, 18, 2766-2778.
Pereoglou, F., Population processes in an early successional heathland 129 Chapter 7: Consolidated references
Gauffre, B., Estoup, A., Bretagnolle, V. & Cosson, J. F. (2008) Spatial genetic structure of a small rodent in a heterogeneous landscape. Molecular Ecology, 17, 4619-4629. Gill, A. M. (1975) Fire and the Australian flora: a review. Australian Forestry, 38, 4-25. Gill, A. M. & Allan, G. (2008) Large fires, fire effects and the fire-regime concept. International Journal of Wildland Fire, 17, 688-695. Gill, A. M. & Catling, P. C. (2002) Fire regimes and biodiversity of forested landscapes of southern Australia. Flammable Australia: The Fire Regimes and Biodiversity of a Continent (eds R. A. Bradstock, J. E. Williams & A. M. Gill), pp. 94-124. Cambridge University Press, New York. Gill, A. M. & Ingwersen, F. (1976) Growth of Xanthorrhoea australis R.Br. in relation to fire. Journal of Applied Ecology, 13, 195-203. Goslee, S. C. & Urban, D. L. (2007) The ecodist package for dissimilarity-based analysis of ecological data. Journal of Statistical Software, 22, 1-19. Goudet, J. (2001) FSTAT, a program to estimate and test gene diversities and fixation indices (version 2.9.3). Available from http://www.unil.ch/izea/softwares/fstat.html. Goudet, J., Perrin, N. & Waser, P. (2002) Tests for sex-biased dispersal using bi-parentally inherited genetic markers. Molecular Ecology, 11, 1103-1114. Gray, S. J., Hurst, J. L., Stidworthy, R., Smith, J., Preston, R. & MacDougall, R. (1998) Microhabitat and spatial dispersion of the grassland mouse (Mus spretus Lataste). Journal of Zoology, 246, 299-308. Greenwood, P. J. (1980) Mating systems, philopatry and dispersal in birds and mammals. Animal Behaviour, 28, 1140-1162. Gros, A., Hovestadt, T. & Poethke, H. J. (2008) Evolution of sex-biased dispersal: The role of sex-specific dispersal costs, demographic stochasticity, and inbreeding Ecological Modelling, 219, 226-233. Guillot, G., Mortier, F. & Estoup, A. (2005) Geneland: a computer package for landscape genetics. Molecular Ecology Notes, 5, 712-715. Guillot, G. & Rousset, F. (2012) On the use of simple and partial Mantel tests in presence of spatial auto-correlation. arXiv:1112.0651v1. Gustafson, E. J. & Gardner, R. H. (1996) The effect of landscape heterogeneity on the probability of patch colonization. Ecology, 77, 94-107. Haering, R. & Fox, B. J. (1995) Habitat utilization patterns of sympatric populations of Pseudomys gracilicaudatus and Rattus lutreolus in coastal heathland: a multivariate analysis. Australian Journal of Ecology, 20, 427-441. Haim, A. & Rozenfeld, F. M. (1995) Temporal segregation in coexisting Acomys species: the possible role of nest site. Journal of Arid Environments, 29, 505-509.
Pereoglou, F., Population processes in an early successional heathland 130 Chapter 7: Consolidated references
Hall, L. S. & Morrison, M. L. (1997) Den and relocation site characteristics and home ranges of Peromyscus truei in the White Mountains of California. Great Basin Naturalist, 57, 124-130. Hanski, I. (2009) Metapopulations and spatial population processes. The Princeton guide to ecology (ed S. A. Levin), pp. 177-185. Princeton University Press, New Jersey. Happold, M. (1976) Social behavior of conilurine rodents (Muridae) of Australia. Zeitschrift Fur Tierpsychologie-Journal of Comparative Ethology, 40, 113-182. Hartung, T. G. & Dewsbury, D. A. (1979) Nest-building behavior in seven species of muroid rodents. Behavioral and Neural Biology, 27, 532-539. Higgs, P. & Fox, B. J. (1993) Interspecific competition: a mechanism for rodent succession after fire in wet heathland. Australian Journal of Ecology, 18, 193-201. Hinkelman, T. M. & Loeb, S. C. (2007) Effect of woody debris abundance on daytime refuge use by cotton mice. Southeastern Naturalist, 6, 393-406. Holland, G. J. & Bennett, A. F. (2010) Habitat fragmentation disrupts the demography of a widespread native mammal. Ecography, 33, 841-853. Hutto, R. L. (1995) Composition of bird communities following stand-replacement fires in northern Rocky Mountain (USA) conifer forests. Conservation Biology, 9, 1041-1058. Hutto, R. L. (2008) The ecological importance of severe wildfires: some like it hot. Ecological Applications, 18, 1827-1834. Ingwersen, F. (1977) Vegetation development after fire in the Jervis Bay Territory. The Australian National University, Canberra. Ives, A. R. (2009) Density dependence and single species population dynamics. The Princeton guide to ecology (ed S. A. Levin), pp. 166-171. Princeton University Press, New Jersey. Jaquiery, J., Broquet, T., Hirzel, A. H., Yearsley, J. & Perrin, N. (2011) Inferring landscape effects on dispersal from genetic distances: how far can we go? Molecular Ecology, 20, 692-705. Keith, D. A., McCaw, W. L. & Whelan, R. J. (2002) Fire regimes in Australian heathlands and their effects on plants and animals. Flammable Australia: the fire regimes and biodiversity of a continent (eds R. A. Bradstock, J. E. Williams & A. M. Gill), pp. 199- 237. Cambridge University Press, New York. Kemper, C. M. (1981) Description of Pseudomys novaehollandiae burrows located with radioisotopes. Australian Mammalogy, 4, 141-3. Klein, H. G. & Layne, J. N. (1978) Nesting behaviour in four species of mice. Journal of Mammalogy, 59, 103-108. Kraaijeveld-Smit, F. J. L., Lindenmayer, D. B., Taylor, A. C., MacGregor, C. & Wertheim, B. (2007) Comparative genetic structure reflects underlying life histories of three sympatric small mammal species in continuous forest of south-eastern Australia. Oikos, 116, 1819-1830.
Pereoglou, F., Population processes in an early successional heathland 131 Chapter 7: Consolidated references
Kraaijeveld-Smit, F. J. L., Ward, S. J. & Temple-Smith, P. D. (2002) Multiple paternity in a field population of a small carnivorous marsupial, the agile antechinus, Antechinus agilis. Behavioural Ecology and Sociobiology, 52, 84-91. Landguth, E. L., Fedy, B. C., Oyler-McCance, S. J., Garey, A. L., Emel, S. L., Mumma, M., Wagner, H. H., Fortin, M.-J. & Cushman, S. A. (2012) Effects of sample size, number of markers, and allelic richness on the dietection of spatial genetic pattern. Molecular Ecology Resources, 12, 276-284. Lebreton, J.-D., Burnham, K. P., Clobert, J. & Anderson, D. R. (1992) Modeling survival and testing biological hypotheses using marked animals: a unified approach with case studies. Ecological Monographs, 62, 67-118. Lee, A. K., Baverstock, P. R. & Watts, C. H. S. (1981) Rodents - the late invaders. Ecological Biogeography of Australia (ed A. Keast), pp. 1521-1553. Dr W Junk Publishers, The Hague, London. Lee, Y., Nelder, J. A. & Pawitan, Y. (2006) Generalized linear models with random effects: unified analysis via h-likelihood, Chapman & Hall/CRC, Boca Raton, FL. Letnic, M. (2003) The effects of experimental patch burning and rainfall on small mammals in the Simpson Desert, Queensland. Wildlife Research, 30, 547-563. Letnic, M. & Dickman, C. R. (2005) The responses of small mammals to patches regenerating after fire and rainfall in the Simpson Desert, central Australia. Austral Ecology, 30, 24- 39. Letnic, M., Dickman, C. R., Tischler, M. K., Tamayo, B. & Beh, C. L. (2004) The responses of small mammals and lizards to post-fire succession and rainfall in arid Australia. Journal of Arid Environments, 59, 85-114. Letnic, M., Tischler, M. & Gordon, C. (2013) Desert small mammal responses to wildfire and predation in the aftermath of a La Nina driven resource pulse. Austral Ecology, In press. Levin, S., A. . (1992) The problem of pattern and scale in ecology: the Robert H. MacArthur award lecture. Ecology, 73, 1943-1967. Lichstein, J. (2007) Multiple regression on distance matrices: a multivariate spatial analysis tool. Plant Ecology, 188, 117-131. Lima, M., Julliard, R., Stenseth, N. & Jaksic, F. M. (2001) Demographic dynamics of a neotropical small rodent (Phyllotis darwini): feedback structure, predation and climatic factors. Journal of Animal Ecology, 70, 761-775. Lindenmayer, D. B., Cunningham, R. B. & Peakall, R. (2005) The recovery of populations of bush rat Rattus fuscipes in forest fragments following major population reduction. Journal of Applied Ecology, 42, 649-658. Lindenmayer, D. B. & Fischer, J. (2006) Habitat fragmentation and landscape change, Island Press, Washington DC.
Pereoglou, F., Population processes in an early successional heathland 132 Chapter 7: Consolidated references
Lindenmayer, D. B., MacGregor, C., Welsh, A., Donnelly, C., Crane, M., Michael, D., Montague-Drake, R., Cunningham, R. B., Brown, D., Fortescue, M., Dexter, N., Hudson, M. & Gill, A. M. (2008a) Contrasting mammal responses to vegetation type and fire. Wildlife Research, 35, 395-408. Lindenmayer, D. B., MacGregor, C., Wood, J. T., Cunningham, R. B., Crane, M., Michael, D., Montague-Drake, R., Brown, D., Fortescue, M., Dexter, N., Hudson, M. & Gill, A. M. (2009) What factors influence rapid post-fire site re-occupancy? A case study of the endangered eastern bristlebird in eastern Australia. International Journal of Wildland Fire, 18, 84-95. Lindenmayer, D. B. & Peakall, R. (2000) The Tumut experiment - integrating demographic and genetic studies to unravel fragmentation effects: a case study of the native bush rat. Genetics, demography and the viability of fragmented populations (eds A. Young & G. Clarke), pp. 173-201. Cambridge University Press, Cambridge. Lindenmayer, D. B., Welsh, A., Donnelly, C. F. & Meggs, R. A. (1996) Use of nest trees by the mountain brushtail possum (Trichosurus caninus) (Phalangeridae: Marsupialia) I. Number of occupied trees and frequency of tree use. Wildlife Research, 23, 343-361. Lindenmayer, D. B., Wood, J. T., MacGregor, C., Michael, D. R., Cunningham, R. B., Crane, M., Montague-Drake, R., Brown, D., Muntz, R. & Driscoll, D. A. (2008b) How predictable are reptile responses to wildfire? Oikos, 117, 1086-1097. Lorenz, G. C. & Barrett, G. W. (1990) Influence of simulated landscape corridors on house mouse (Mus musculus) dispersal. American Midland Naturalist, 123, 348-356. Lunney, D. (2008a) Bush rat, Rattus fuscipes. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 685-687. Reed New Holland, Sydney, Australia. Lunney, D. (2008b) Swamp rat, Rattus lutreolus. The Mammals of Australia (eds S. Van Dyck & R. Strahan), pp. 690-692. Reed New Holland, Sydney, Australia. Luo, J. & Fox, B. J. (1994) Diet of the Eastern chestnut mouse (Pseudomys gracilicaudatus) II Seasonal and successional patterns. Wildlife Research, 21, 419-431. Luo, J. & Fox, B. J. (1995) Competitive effects of Rattus lutreolus presence on food resource use by Pseudomys gracilicaudatus. Australian Journal of Ecology, 20, 556-564. Luo, J., Fox, B. J. & Jefferys, E. (1994) Diet of the Eastern Chestnut Mouse (Pseudomys gracilicaudatus) I Composition, diversity and individual variation. Wildlife Research, 21, 401-417. Luo, J. I. A. & Fox, B. J. (1996) Seasonal and successional dietary shifts of two sympatric rodents in coastal heathland: A possible mechanism for coexistence. Austral Ecology, 21, 121-132. MacArthur, R. H. & Wilson, E. O. (1967) The Theory of Island Biogeography, Princeton University Press, New Jersey.
Pereoglou, F., Population processes in an early successional heathland 133 Chapter 7: Consolidated references
Macqueen, P. E., Nicholls, J. A., Hazlitt, S. L. & Goldizen, A. W. (2008) Gene flow among native bush rat, Rattus fuscipes (Rodentia : Muridae), populations in the fragmented subtropical forests of south-east Queensland. Austral Ecology, 33, 585-593. Madison, D. M., Hill, J. P. & Gleason, P. E. (1984) Seasonality in the nesting behavior of Peromyscus leucopus. American Midland Naturalist, 112, 201-204. Mahoney, J. A. & Posamentier, H. (1975) The occurrence of the native rodent, Pseudomys gracilicaudatus (Gould, 1845) [Rodentia: Muridae], in New South Wales. Australian Mammalogy, 1, 333-346. Manel, S., Schwartz, M. K., Luikart, G. & Taberlet, P. (2003) Landscape genetics: combining landscape ecology and population genetics. Trends in Ecology & Evolution, 18, 189- 197. Masters, P. (1993) The effects of fire driven succession and rainfall on small mammals in spinifex grassland at Uluru National Park, Northern Territory. Wildlife Research, 20, 803-813. McCarthy, M. A., Gill, A. M. & Lindenmayer, D. B. (1999) Fire regimes in mountain ash forest: evidence from forest age structure, extinction models and wildlife habitat. Forest Ecology and Management, 124, 193-203. McCullagh, P. & Nelder, J. A. (1989) Generalized linear models, Chapman and Hall, New York. McCulloch, C. E., Searle, S. R. & Neuhaus, J. M. (2008) Generalized, linear, and mixed models, Wiley, Hoboken, N.J. . McDonough, T. J. & Rexstad, E. (2005) Short-term demographic response of the red-backed vole to spruce beetle infestations in Alaska. Journal of Wildlife Management, 69, 246- 254. McRae, B. H. & Beier, P. (2007) Circuit theory predicts gene flow in plant and animal populations. Proceedings of the National Academy of Sciences of the United States of America, 104, 19885-19890. McRae, B. H., Dickson, B. G., Keitt, T. H. & Shah, V. B. (2008) Using circuit theory to model connectivity in ecology, evolution and conservation. Ecology, 89, 2712-2724. McRae, B. H. & Shah, V. B. (2009) Circuitscape User Guide. University of California, Santa Barbara. Available at http://www.circuitscape.org. Meek, P. D. & Triggs, B. (1997) Extension to the range of the eastern chestnut mouse, Pseudomys gracilicaudatus, to Jervis Bay, New South Wales. Australian Mammalogy, 20, 123-125. Miller, A. D., Roxburgh, S. H. & Shea, K. (2011) How frequency and intensity shape diversity- disturbance relationships. Proceedings of the National Academy of Sciences, 108, 5643- 5648.
Pereoglou, F., Population processes in an early successional heathland 134 Chapter 7: Consolidated references
Mills, L. S. (2013) Conservation of wildlife populations: demography, genetics and management. 2nd Edn, John Wiley and Sons, Ltd. Monamy, V. & Fox, B. J. (2000) Small mammal succession is determined by vegetation density rather than time elapsed since disturbance. Austral Ecology, 25, 580-587. Monamy, V. & Fox, B. J. (2010) Responses of two species of heathland rodents to habitat manipulation: vegetation density thresholds and the habitat accommodation model. Austral Ecology, 35, 334-347. Moro, D. & Morris, K. (2000) Movements and refugia of Lakeland Downs short-tailed mice, Leggadina lakedownensis, and house mice, Mus domesticus, on Thevenard Island, Western Australia. Wildlife Research, 27, 11-20. Moro, D. & Spencer, P. B. S. (2003) Microsatellite primers for the western pebble-mound mouse (Pseudomys chapmani) that show cross amplification for other species of Australian rodent. Molecular Ecology Notes, 3, 259-261. Morris, D. W. (1992) Scales and costs of habitat selection in heterogeneous landscapes. Evolutionary Ecology, 6, 412-432. Morris, D. W., Fox, B. J., Luo, J. & Monamy, V. (2000) Habitat-dependent competition and the coexistence of Australian heathland rodents. Oikos, 91, 294-306. Morrison, M. L., Marcot, B. G. & Mannan, R. W. (2006) Wildlife-habitat relationships: concepts and applications, Island Press, Washington. Morzillo, A. T., Feldhamer, G. A. & Nicholson, M. C. (2003) Home range and nest use of the golden mouse (Ochrotomys nuttalli) in Southern Illinois. Journal of Mammalogy, 84, 553-560. Mossman, C. A. & Waser, P. M. (2001) Effects of habitat fragmentation on population genetic structure in the white-footed mouse (Peromyscus leucopus). Canadian Journal of Zoology-Revue Canadienne De Zoologie, 79, 285-295. Murphy, M. A., Dezzani, R., Pilliod, D. S. & Storfer, A. (2010) Landscape genetics of high mountain frog metapopulations. Molecular Ecology, 19, 3634-3649. Newsome, A. E. & Catling, P. C. (1983) Animal demography in relation to fire and shortage of food: some indicative models. Ecological Studies, 43, 490-505. Newsome, A. E., McIlroy, J. & Catling, P. C. (1975) The effects of an extensive wildfire on populations of twenty ground vertebrates in southeast Australia. Proceedings of the Ecological Society of Australia, 9, 107-123. Nupp, T. E. & Swihart, R. K. (1996) Effect of forest patch area on population attributes of white-footed mice (Peromyscus leucopus) in fragmented landscapes. Canadian Journal of Zoology-Revue Canadienne De Zoologie, 74, 467-472. Ostfeld, R. S. (1985) Limiting resources and territoriality in microtine rodents. American Naturalist, 126, 1-15.
Pereoglou, F., Population processes in an early successional heathland 135 Chapter 7: Consolidated references
Paetkau, D., Slade, R., Burden, M. & Estoup, A. (2004) Direct real-time estimation of migration rate using assignment methods: a simulation-based exploration of accuracy and power. Molecular Ecology, 13, 55-65. Paradis, E., Guedon, G. & Pradel, R. (1993) Estimation of sex- and age-related survival rates in a microtine population. Journal of Wildlife Management, 57, 158-163. Pardon, L. G., Brook, B. W., Griffiths, A. D. & Braithwaite, R. W. (2003) Determinants of survival for the northern brown bandicoot under a landscape-scale fire experiment. Journal of Animal Ecology, 72, 106-115. Park, S. D. E. (2001) Trypanotolerance in West African Cattle and the Population Genetic Effects of Selection. University of Dublin, Dublin, Ireland. Pavlacky, D. C., Goldizen, A. W., Prentis, P. J., Nicholls, J. A. & Lowe, A. J. (2009) A landscape genetics approach for quantifying the relative influence of historic and contemporary habitat heterogeneity on the genetic connectivity of a rainforest bird. Molecular Ecology, 18, 2945-2960. Peakall, R., Ruibal, M. & Lindenmayer, D. B. (2003) Spatial autocorrelation analysis offers new insights into gene flow in the Australian bush rat, Rattus fuscipes. Evolution, 57, 1182- 1195. Peakall, R. & Smouse, P. E. (2006) GenAlEx 6: genetic analysis in Excel. Population genetic software for teaching and research. Molecular Ecology Notes, 6, 288-295. Pereoglou, F., Lindenmayer, D. B., MacGregor, C., Ford, F., Wood, J. & Banks, S. C. (2013) Landscape genetics of an early successional specialist in a disturbance-prone environment. Molecular Ecology, 22, 1267-1281. Pereoglou, F., MacGregor, C., Banks, S. C., Ford, F. D., Wood, J. T. & Lindenmayer, D. B. (2011) Refuge site selection by the eastern chestnut mouse in recently burned heath. Wildlife Research, 38, 290-298. Perez-Espona, S., Perez-Barberia, F. J., McLeod, J. E., Jiggins, C. D., Gordon, I. J. & Pemberton, J. M. (2008) Landscape features affect gene flow of Scottish Highland red deer (Cervus elaphus). Molecular Ecology, 17, 981-996. Peterson, G. D. (2002) Contagious disturbance, ecological memory, and the emergence of landscape pattern. Ecosystems, 5, 329-338. Pickett, S. T. A. & White, P. S. (1985) The Ecology of Natural Disturbance and Patch Dynamics. Academic Press Inc, Florida. Pierson, J. C., Allendorf, F. W., Drapeau, P. & Schwartz, M. K. (2013) Breed locally, disperse globally: fine-scale genetic structure despite landscape-scale panmixia in a fire- specialist. PLoS ONE, 8, e67248. Piry, S., Alapetite, A., Cornuet, J. M., Paetkau, D., Baudouin, L. & Estoup, A. (2004) GeneClass2: A software for genetic assignment and first-generation migrant detection. Journal of Heredity, 95, 536-539.
Pereoglou, F., Population processes in an early successional heathland 136 Chapter 7: Consolidated references
Pocock, M. J. O., Searle, J. B. & White, P. C. L. (2004) Adaptations of animals to commensal habitats: population dynamics of house mice Mus musculus domesticus on farms. Journal of Animal Ecology, 73, 878-888. Pons, P., Henry, P.-Y., Gargallo, G., Prodon, R. & Lebreton, J.-D. (2003) Local survival after fire in Mediterranean shrublands: combining capture-recapture data over several bird species. Population Ecology, 45, 187-196. Pradel, R. (1996a) Utilization of capture-mark-recapture for the study of recruitment and population growth rate. Biometrics, 52, 703-709. Pradel, R. (1996b) Utilization of capture-mark-recapture for the study of recruitment and population growth rate. Biometrics, 52, 703-709. Pritchard, J. K., Stephens, M. & Donnelly, P. (2000) Inference of population structure using multilocus genotype data. Genetics, 155, 945-959. Prugh, L. R., Hodges, K. E., Sinclair, A. R. E. & Brashares, J. S. (2008) Effect of habitat area and isolation on fragmented animal populations. Proceedings of the National Academy of Sciences, 105, 20770-20775. Pryde, M. A., O'Donnell, C. F. J. & Barker, R. J. (2005) Factors influencing survival and long- term population viability of New Zealand long-tailed bats (Chalinolobus tuberculatus): implications for conservation. Biological Conservation, 126, 175-185. Rannala, B. & Mountain, J. L. (1997) Detecting immigration using multilocus genotypes. Proceedings of the National Academy of Sciences, 94, 9197-9221. RDevelopmentCoreTeam (2008) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. Robinson, A. C. (1987) The ecology of the bush rat, Rattus fuscipes (Rodentia: Muridae), in Sherbrooke Forest, Victoria. Australian Mammalogy, 11, 35-49. Robinson, N. M., Leonard, S. W. J., Ritchie, E. G., Bassett, M., Chia, E. K., Buckingham, S., Gibb, H., Bennett, A. F. & Clarke, M. F. (2013) Refuges for fauna in fire-prone landscapes: their ecological function and importance. Journal of Applied Ecology, 50, 1321-1329. Rossi, F., Forster, R. M., Montserrat, F., Ponti, M., Terlizzi, A., Ysebaert, T. & Middelburg, J. J. (2007) Human trampling as short-term disturbance on intertidal mudflats: effects on macrofauna biodiversity and population dynamics of bivalves. Marine Biology, 151, 2077-2090. Rota, C. T., Millspaugh, J. J., Rumble, M. A., Lehman, C. P. & Kesler, D. C. (2014) The role of wildfire, prescribed fire, and mountain pine beetle infestations on the population dynamics of black-backed woodpeckers in the Black Hills, South Dakota. PLoS ONE, 9, e94700.
Pereoglou, F., Population processes in an early successional heathland 137 Chapter 7: Consolidated references
Ruscoe, W. A., Mather, P. B. & Wilson, J. (1998) Genetic structure in populations of Rattus sordidus in sugarcane-growing districts of Queensland, Australia. Journal of Mammalogy, 79, 612-623. Sandercock, B. K. (2006) Estimate of demographic parameters from live-encounter data: a summary review. Journal of Wildlife Management, 70, 1504-1520. Sanecki, G. M., Green, K., Wood, H., Lindenmayer, D. & Sanecki, K. L. (2006) The influence of snow cover on home range and activity of the bush-rat (Rattus fuscipes) and the dusky antechinus (Antechinus swainsonii). Wildlife Research, 33, 489-496. Schaub, M., Jakober, H. & Stauber, W. (2013) Strong contribution of immigration to local population regulation: evidence from a migratory passerine. Ecology, 94, 1828-1838. Schaub, M., Reichlin, T. S., Abadi, F., Kery, M., Jenni, L. & Arlettaz, R. (2012) The demographic drivers of local population dynamics in two rare migratory birds. Oecologia, 168, 97-108. Schrey, A. W., Ashton, K. G., Heath, S., McCoy, E. D. & Mushinsky, H. R. (2011a) Fire alters patterns of genetic diversity among three lizard species in Florida Scrub habitat. J. Hered. , 102, 399-408. Schrey, A. W., Fox, A. M., Mushinsky, H. R. & McCoy, E. D. (2011b) Fire increases variance in genetic characteristics of Florida sand skink (Plestiodon reynoldsi) local populations. Molecular Ecology, 20, 56-66. Schuelke, M. (2000) An economic method for the fluorescent labeling of PCR fragments. Nature Biotechology, 18, 233-234. Schultz, C. B. (1998) Dispersal behavior and its implications for reserve design in a rare oregon butterfly. Conservation Biology, 12, 284-292. Schwarz, G. (1978) Estimating the dimension of a model. Annals of Statistics, 6, 461-464. Scribner, K. T., Blanchong, J. A., Bruggeman, D. J., Epperson, B. K., Lee, C.-Y., Pan, Y.-W., Shorey, R. I., Prince, H. H., Winterstein, S. R. & Luukkonen, D. R. (2005) Geographical genetics: conceptual foundations and empirical applications of spatial genetic data in wildlife management. Journal of Wildlife Management, 69, 1434-1453. Selonen, V. & Hanski, I. K. (2004) Young flying squirrels (Pteromys volans) dispersing in fragmented forests. Behavioral Ecology, 15, 564-571. Short Bull, R. A., Cushman, S. A., Mace, R., Chilton, T., Kendall, K. C., Landguth, E. L., Schwartz, M. K., McKelvey, K., Allendorf, F. W. & Luikart, G. (2011) Why replication is important in landscape genetics: American black bear in the Rocky Mountains. Molecular Ecology, 20, 1092-1107. Sibly, R. M., Collett, D., Promislow, D. E. L., Peacock, D. J. & Harvey, P. H. (1997) Mortality rates of mammals. Journal of Zoology, London, 243, 1-12. Smouse, P. E. & Peakall, R. (1999) Spatial autocorrelation analysis of individual multiallele and multilocus genetic structure. Heredity, 82, 561-573.
Pereoglou, F., Population processes in an early successional heathland 138 Chapter 7: Consolidated references
Smouse, P. E., Peakall, R. & Gonzales, E. (2008) A heterogeneity test for fine-scale genetic structure. Molecular Ecology, 17, 3389-3400. Soga, M. & Koike, S. (2013) Large forest patches promote breeding success of a terrestrial mammal in urban landscapes. PLoS ONE, 8, e51802. Sousa, W. P. (1979) Disturbance in intertidal boulder fields: the non-equilibrium maintenance of species diversity. Ecology, 60, 1225-1239. Spear, S. F., Balkenhol, N., Fortin, M.-J., McRae, B. H. & Scribner, K. (2010) Use of resistance surfaces for landscape genetic studies: considerations for parameterization and analysis. Molecular Ecology, 19, 3576-3591. Spear, S. F., Peterson, C. R., Matocq, M. D. & Storfer, A. (2005) Landscape genetics of the blotched tiger salamander (Ambystoma tigrinum melanostictum). Molecular Ecology, 14, 2553-2564. Spear, S. F. & Storfer, A. (2008) Landscape genetic structure of coastal tailed frogs (Ascaphus truei) in protected vs. managed forests. Molecular Ecology, 17, 4642-4656. Spear, S. F. & Storfer, A. (2010) Anthropogenic and natural disturbance lead to differing patterns of gene flow in the Rocky Mountain tailed frog, Ascaphus montanus. Biological Conservation, 143, 778-786. Specht, R. L. (1981) Responses to fire in heathlands and related shrublands. Fire and the Australian Biota (eds A. M. Gill, R. H. Groves & I. R. Noble), pp. 395-415. Australian Academy of Science, Canberra. Steffen, W., Burbidge, A. A., Hughes, L., Kitching, R., Lindenmayer, D. B., Musgrave, W., Stafford Smith, M. & Werner, P. (2009) Australia's Biodiversity and Climate Change, CSIRO Publishing, Melbourne. Stokes, M. K. (1995) Selection of refuge sites by sympatric Microtus ochrogaster and Sigmodon hispidus. Journal of Mammalogy, 76, 83. Storfer, A., Murphy, M. A., Spear, S. F., Holderegger, R. & Waits, L. P. (2010) Landscape genetics: where are we now? Molecular Ecology, 19, 3496-3514. Stow, A. J. & Magnusson, W. E. (2012) Genetically defining populations is of limited use for evaluating and managing human impacts on gene flow. Wildlife Research, 39, 290-294. Strahan, R. (1995) The Mammals of Australia. Reed, Chatswood. Strahan, R. (2008) The Mammals of Australia. Reed, Sydney. Sutherland, E. F. & Dickman, C. R. (1999) Mechanisms of recovery after fire by rodents in the Australian environment: a review. Wildlife Research, 26, 405-419. Swanson, M. E., Franklin, J. F., Beschta, R. L., Crisafulli, C. M., DellaSala, D. A., Hutto, R. L., Lindenmayer, D. B. & Swanson, F. J. (2011) The forgotten stage of forest succession: early-successional ecosystems on forest sites. Frontiers in Ecology and the Environment, 9, 117-125.
Pereoglou, F., Population processes in an early successional heathland 139 Chapter 7: Consolidated references
Taws, N. (1997) Vegetation survey and mapping of Jervis Bay Territory. Report to Environment Australia. Canberra. Taylor, J. M. & Horner, B. E. (1972) Observations on reproductive biology of Pseudomys (Rodentia: Muridae). Journal of Mammalogy, 53, 318-328. Taylor, P. D., Fahrig, L., Henein, K. & Merriam, G. (1993) Connectivity is a vital element of landscape structure. Oikos, 68, 571-573. Templeton, A. R., Brazeal, H. & Neuwald, J. L. (2011) The transition from isolated patches to a metapopulation in the eastern collared lizard in response to prescribed fires. Ecology, 92, 1736-1747. Templeton, A. R., Robertson, R. J., Brisson, J. & Strasburg, J. (2001) Disrupting evolutionary processes: the effect of habitat fragmentation on collared lizards in the Missouri Ozarks. Proceedings of the National Academy of Sciences, 98, 5426-5432. Thomas, C. D. (1994) Extinction, colonization, and metapopulations: environmental tracking by rare species. Conservation Biology, 8, 373-378. Thomas, C. D., Thomas, J. A. & Warren, M. S. (1992) Distributions of occupied and vacant butterfly habitats in fragmented landscapes. Oecologia, 92, 563-567. Thompson, M. B., Medlin, G., Hutchinson, R. & West, N. (1989) Short-term effects of fuel reduction burning on populations of small terrestrial mammals. Australian Wildlife Research, 16, 117-129. Thompson, P. T. & Fox, B. J. (1993) Asymmetric competition in Australian heathland rodents: a reciprocal removal experiment demonstrating the influence of size-class structure. Oikos, 67, 264-278. Thornton, D. H., Branch, L. C. & Sunquist, M. E. (2011) The influence of landscape, patch, and with-in patch factors on species presence and abundance: a review of focal patch studies. Landscape Ecology, 26, 7-18. Townley, S. (2007) Targeted survey for the eastern chestnut mouse Pseudomys gracilicaudatus, New Holland mouse Pseudomys novaehollandiae, eastern pygmy-possum Cercartetus nanus, and long-nosed bandicoot Perameles nasuta on Pacific City Estate lands, Jervis Bay National Park. NSW Department of Environment. Townley, S. J. (2000) The ecology of the Hastings River mouse Pseudomys oralis (Rodentia: Muridae) in northeastern New South Wales and southeastern Queensland. Southern Cross University, Lismore, New South Wales, Australia. Troyer, E. M., Cameron Devitt, S. E., Sunquist, M. E., Goswami, V. R. & Oli, M. K. (2014) Density dependence or climatic variation? Factors influencing survival, recruitment, and population growth rate of Virginia opossums. Journal of Mammalogy, 95, 421-430. Turner, M. G. (1989) Landscape ecology: the effect of pattern on process. Annual Review of Ecology and Systematics, 20, 171-197.
Pereoglou, F., Population processes in an early successional heathland 140 Chapter 7: Consolidated references
Turner, M. G. (2005) Landscape ecology: what is the state of the science. Annual Review of Ecology and Systematics, 36, 319-344. Turner, M. G., Baker, W. L., Peterson, C. J. & Peet, R. K. (1998) Factors influencing succession: lessons from large, infrequent natural disturbances. Ecosystems, 1, 511-523. Umapathy, G. & Kumar, A. (2000) The demography of the lion-tailed macaque (Macaca silenus) in rain forest fragments in the Anamalai Hills, South India. Primates, 41, 119- 126. Velle, L. G. & Vandvik, V. (2014) Succession after prescribed burning in coastal Calluna heathlands along a 340-km latitudinal gradient. Journal of Vegetative Science, 25, 546- 558. Vieira, E. M. & Briani, D. C. (2013) Short-term effects of fire on small rodents in the Brazillian Cerrad and their relation with feeding habits. International Journal of Wildland Fire, In press. Villafuerte, R., Litvaitis, J. A. & Smith, D. F. (1997) Physiological responses by lagomorphs to resource limitations imposed by habitat fragmentation: implications for condition- sensitive predation. Canadian Journal of Zoology-Revue Canadienne De Zoologie, 75, 148-151. Villella, R. F., Smith, D. R. & Lemarie, D. P. (2004) Estimating survival and recruitment in a freshwater mussel population using mark-recapture technique. The American Midland Naturalist, 151, 114-133. Wade, M. J. & McCauley, D. E. (1988) Extinction and recolonisation: their effects on the genetic differentiation of local populations. Evolution, 42, 995-1005. Wagner, D. M., Feldhamer, G. A. & Newman, J. A. (2000) Microhabitat selection by golden mice (Ochrotomys nuttalli) at arboreal nest sites. American Midland Naturalist, 144, 220-225. Wang, Y.-H., Yang, K.-C., Bridgman, C. & Lin, L.-K. (2008) Habitat suitability modeling to correlate gene flow with landscape connectivity. Landscape Ecology, 23, 989–1000. Ward, G. D. (1978) Habitat use and home range of radio-tagged opossums Trichosurus vulpecula (Kerr) in New Zealand lowland forest. The Ecology of Arboreal Folivores (ed G. G. Montogomery), pp. 305-313. Smithsonian Institute Press, Washington DC. Warneke, R. M. (1971) Field study of the bush rat (Rattus fuscipes). Wildlife Contributions Victoria, 14, 1-115. Watson, S. J., Taylor, R. S., Nimmo, D. G., Kelly, L. T., Clarke, M. F. & Bennett, A. F. (2012) The influence of unburnt patches and distance from refuges on post-fire bird communities. Animal Conservation, 15, 499-507. Watts, C. H. S. & Aslin, H. J. (1981) The Rodents of Australia, Angus & Robertson Publishers, Sydney.
Pereoglou, F., Population processes in an early successional heathland 141 Chapter 7: Consolidated references
Watts, C. H. S. & Kemper, C. M. (1989) Muridae. Fauna of Australia, pp. 1-35. AGPS, Canberra. West, S. D. & Dublin, H. T. (1984) Behavioural strategies of small mammals under winter conditions: solitary or social? Proceedings of the International Colloquium: Winter Ecology of Small Mammals (ed J. F. Merritt), pp. 293-299. Carnegie Museum of Natural History, Pittsburgh. Whelan, R. J. (1995) The Ecology of Fire, Cambridge University Press, Cambridge. Whelan, R. J., Rodgerson, L., Dickman, C. R. & Sutherland, E. F. (2002) Critical life cycles of plants and animals: developing a process-based understanding of population changes in fire-prone landscapes. Flammable Australia: The Fire Regimes and Biodiversity of a Continent (eds R. A. Bradstock, J. E. Williams & A. M. Gill), pp. 94-124. Cambridge University Press, New York. White, G. C. & Burnham, K. P. ( 1999) Program MARK: Survival estimation from populations of marked animals. Bird Study, 46 Supplement, 120-138. Williams, A. A. J., Karoly, D. J. & Tapper, N. (2001) The sensitivity of Australian fire danger to climate change. Climatic Change, 49, 171-191. Williams, D. G. (1995) Heaths and scrubs. Jervis Bay: A Place of Cultural, Scientific and Educational Value. Kowari 5 (eds G. Cho, A. Georges, R. Stoutjesdijk & R. Longmore), pp. 83-90. Australian Nature Conservation Agency, Canberra. Wolfe, J. L. (1970) Experiments on nest-building behaviour in Peromyscus (Rodentia: Cricetinae). Animal Behaviour, 18, 613-615. Wolff, J. O. & Hurlbutt, B. (1982) Day refuges of Peromyscus leucopus and Peromyscus maniculatus. Journal of Mammalogy, 63, 666-668. Wolton, R. J. & Trowbridge, B. J. (1985) The ranging and nesting behavior of wood mice, Apodemus sylvaticus (Rodentia: Muridae), as revealed by radio-tracking. Journal of Zoology, 206, 203-224. Woods, R. E. & Ford, F. D. (2000) Observations on the behaviour of the smoky mouse Pseudomys fumeus (Rodentia: Muridae). Australian Mammalogy, 22, 35-42. Wright, S. (1943) Isolation by distance. Genetics, 28, 139-156.
Pereoglou, F., Population processes in an early successional heathland 142
Appendix 1: Manuscript – Refuge site selection
Manuscript: Refuge site selection in recently burned heath Published, September 2011
Pereoglou, F., Population processes in an early successional heathland 143 Appendix 1:Manuscript - Refuge site selection
CSIRO PUBLISHING Wildlife Research, 2011, 38, 290–298 www.publish.csiro.au/journals/wr
Refuge site selection by the eastern chestnut mouse in recently burnt heath
F. Pereoglou A,C, C. Macgregor A, S. C. Banks A, F. Ford B, J. Wood A and D. B. Lindenmayer A
AFenner School of Environment and Society, Australian National University, Canberra, ACT 0200, Australia. BCSIRO Ecosystem Sciences, Australian National Wildlife Collection, GPO Box 284, Canberra, ACT 2601, Australia. CCorresponding author. Email: [email protected]
Abstract Context. Temporal reduction in shelter is an indirect primary impact of fire. Removal of animal refugia has implications for shelter site selection and fidelity – factors that have been largely overlooked in studies of Australian rodent fauna. This information is critical for guiding species conservation and appropriate land management including prescribed burning practices. Aims. We sought to determine which features of burnt heath were selected as shelter sites by the eastern chestnut mouse, whether there was sex and/or seasonal variation in shelter site selection and whether we could identify primary refugia. Methods. We completed a radio-telemetry study to identify diurnal refuge sites and compare habitat attributes with those of a matched set of control sites. We then used habitat features and fidelity parameters to classify refuge site use. Key results. We found the eastern chestnut mouse selected shelters with the presence of specific structures and had taller, denser vegetation than randomly selected control sites. There were no differences in habitat selection between the sexes. Shelter sites in the non-breeding season had greater vegetation density compared with those used in the breeding season. In the breeding season, the eastern chestnut mouse showed no evidence of increased fidelity to particular refuges. Vegetation density in winter was the best predictor of a primary refuge compared with whether or not an animal returned to a shelter site or the amount of time spent in a shelter site. Mice were ephemeral and non-gregarious in their refuge use. There was some evidence for inheritance of refuge sites from a parent, as well as inter-season shelter site fidelity. Conclusions. The eastern chestnut mouse selected refugia that had habitat attributes offering maximum protection. Seasonality in refuge site selection is likely to reflect the reproductive and environmental trade-offs in critical resources during different seasons. The maintenance of multiple, rarely occupied shelters by the eastern chestnut mouse is consistent with data for other mammals. Implications. Fire management should ensure retention of vegetation structure on the ground layer, dense habitat patches in burned areas, and be carefully planned during the winter season to maintain shelter and refuge sites to assist population persistence.
Introduction 1993; Monamy and Fox 2000; Fox et al. 2003; Monamy and Fox The response of small mammal populations to fire has been 2010). studied extensively in Australia (Newsome et al. 1975; The eastern chestnut mouse occurs at low densities when Cockburn et al. 1981; Fox 1982; Newsome and Catling 1983; present, is rare or extinct in many parts of its range (Breed and Catling 1986; Thompson et al. 1989; Catling 1991; Masters Ford 2007) and is classified as vulnerable under the Threatened 1993; Letnic 2003; Letnic et al. 2004; Lindenmayer et al. Species Conservation Act 1992 (NSW). Inappropriate fire 2008). A pattern of post-fire irruption followed by decline has regimes including fire suppression in areas supporting been observed for some Australian native rodents (Whelan et al. populations of the eastern chestnut mouse may affect the 2002). The eastern chestnut mouse (Pseudomys gracilicaudatus) species’ persistence and viability. However, the mechanisms is one example. Existing evidence suggests the species is a pyric underlying the species’ fire response are unknown. specialist with peak population density in recently burnt Terrestrial small mammals can be affected by fire through heathland and a subsequent decline in abundance with time direct mortality, and indirectly through changes in resource since fire (Fox 1982; Higgs and Fox 1993; Thompson and Fox availability (Whelan et al. 2002). Studies of the eastern
CSIRO 2011 10.1071/WR11007 1035-3712/11/040290
Pereoglou, F., Population processes in an early successional heathland 144 Appendix 1:Manuscript - Refuge site selection
Refuge site selection by the eastern chestnut mouse Wildlife Research 291
chestnut mouse’s dietary requirements indicate the species is a (iii) Is there a primary–secondary model of refuge use? We generalist herbivore with an opportunistic foraging strategy hypothesised that animals would show increased fidelity able to utilise a diverse food resource (Luo and Fox 1994; Luo to shelters characterised by particular attributes including et al. 1994). A diverse diet is a basic requirement for rapid the presence of a mouse-made nesting structure (Wolfe colonisation of recently burnt areas and is an essential attribute 1970; Klein and Layne 1978; Hartung and Dewsbury for a fire opportunist to take advantage of all available food 1979). These primary refuges would be used more resources (Luo et al. 1994). Given this, it is unlikely that frequently and would have habitat attributes that depletion of the food resource alone would be a determinant of distinguish them from secondary shelters. the species’ fire response. (iv) Does the eastern chestnut mouse share shelter sites? We The role of animal refugia in facilitating population recovery expected that individuals would not share shelter sites, and persistence after a fire is poorly understood for Australian instead likely to be solitary in their nesting habits. The rodent fauna. A reduction in available shelter sites is one of the closely related western chestnut mouse (P. nanus) also has a primary impacts of fire on populations (Newsome and Catling dispersed, solitary social organisation (Breed and Ford 1983; Friend 1993) and may have multiple consequences 2007) and other Pseudomys species have been shown to including distributional, demographic and behavioural change engage in agonistic nest defence (Happold 1976). (Sutherland and Dickman 1999). The presence, size and location Identification of small mammal shelter requirements and an of refugia are landscape attributes that may influence the understanding of refuge site fidelity and nesting behaviour could processes of mortality, colonisation, post-fire survival, be used by land managers when designing appropriate burning establishment, reproduction and population increase (Whelan regimes or promoting post-fire population recovery. For et al. 2002). Daytime refugia and shelter sites are critical for threatened species, knowledge of nest requirements surviving disturbance events (Braithwaite 1990), environmental and daytime refuge habitat and use is fundamental for extremes, predation and for providing security while weaning conservation efforts. young (Frank and Layne 1992). Identification of habitat components that might represent refuges and the question of fi how important they are for maintenance of post- re populations Materials and methods requires additional study (Keith et al. 2002). Study sites The rate of reappearance after fire is dependent on a species’ mobility, dispersal capacity, dietary specialisation We selected six sites for study in Booderee and Jervis Bay and habitat preferences (Keith et al. 2002), including the level national parks in south-eastern Australia (Fig. 1). These of protection provided by refugia during and after a fire. adjoining reserves are located ~200 km south of Sydney on the An assumption of survival within burnt areas would lead us to south coast of New South Wales. We based site selection on predict rapid reappearance and population growth (Keith et al. survey data of Townley (2007), Lindenmayer et al.(2008) and a 2002). Recruitment from residual animals after disturbance pilot study indicating the presence of the eastern chestnut mouse has been shown to be important (Lindenmayer et al. 2005; at sufficiently high densities to enable attachment of radio- Banks et al. 2011). This suggests refugia may play a critical transmitters to at least two individuals per site. We classified role in a species’ ability to survive fire and the subsequent study sites as heath (defined by Taws (1997) as vegetation population response. The identification, distribution and dominated by shrubs with small narrow leaves), generally less protection of animal shelter sites and refuges should be an than 2 m tall and varyingly dominated by Banksia ericifolia and essential component of conservation strategies in fire-prone Sprengelia incarnata, with Allocasuarina distyla, Isopogon landscapes. anemonifolius and Hakea teretifolia also being common. Each In this paper, we document the results of a radio-telemetry of our six field sites had been burnt by wildfire within the last study of daytime refuge habitat requirements and nest fidelity of six years. the eastern chestnut mouse. We sought to answer the following questions: Trapping and radio-collaring We completed trapping and radio-collaring during the 2008–09 (i) Does the eastern chestnut mouse preferentially select summer breeding season (November–March) and the winter shelter sites based on specific habitat attributes? We 2009 non-breeding season (June–August). We captured expected that mouse-selected shelter sites would be animals using grids of type A Elliot traps open for four characterised by denser, taller vegetation than control sites. consecutive nights using a bait of rolled oats, peanut butter (ii) Are there sex differences or seasonal variation in shelter site and fennel seed. During winter, we insulated traps with non- selection and fidelity? We hypothesised the female mice absorbent cotton wool and covered them with a plastic sleeve. would select shelter sites that offer optimal cover and Our grid arrangement comprised lines at 20 m intervals with protection during summer and would show greater traps 10 m apart. We provide a summary of field site details, shelter site fidelity than males. We expected these trapping and radio-collaring data (Table 1). differences to be driven by different requirements of the We assessed the weight, sex, age cohort and breeding sexes during the breeding season – care of offspring in condition of captured animals. We marked mice for permanent fi females versus proximity to mates for males. We predicted identi cation using Trovan ID100 implantable transponders there would be no difference between male and female mice (Microchips Australia Ltd, Melbourne). We selected mice for fi in shelter site selection and delity during winter. radio-collar attachment if they were (i) an adult, (ii) weighed
Pereoglou, F., Population processes in an early successional heathland 145 Appendix 1:Manuscript - Refuge site selection
292 Wildlife Research F. Pereoglou et al.
280 000 285 000 290 000 295 000 N 01 2 km 6 115 000 6 115 000
Jervis Bay 6 110 000 6 110 000
Booderee National Park Botanical Gardens 6 105 000 6 105 000 Department of Defence Wreck Bay Aboriginal Land Jervis Bay National Park
Study sites
280 000 285 000 290 000 295 000
Fig. 1. Location of study sites in Booderee and Jervis Bay national parks.
Table 1. Trapping and radio-collaring site summary BNP, Booderee National Park; JNP, Jervis Bay National Park; † average value when effort varied between trap events. The number of captures includes previously marked recaptures. The number in parentheses indicates number of animals radio-collared in the previous season
Location Site Area Trap-nights No. of No. of <,Trap-nights No. of No. of collars <, (ha) summer captures collars fitted winter captures fitted BNP 7–44B 2.40 960 18 4 1 3 480 10 6 (1) 2 4 BNP 7–I 2.05† 820 17 6 4 2 400 8 6 (2) 2 4 BNP 7–2 2.00 400 19 4 3 1 400 13 6 (1) 4 2 BNP 7–46 3.25† 1280 36 7 2 5 480 7 5 (0) 4 1 BNP 7–H 2.20† 880 20 5 2 3 400 5 3 (1) 2 1 JNP 7–ST1 2.50 500 7 3 1 2 500 11 3 (1) 1 2 Total 14.40 4840 117 29 13 16 2660 54 29 (6) 15 14
>70 g to ensure collar weight was less than 5% of an individual’s within 1 m2. We numbered and permanently marked shelter bodyweight, and (iii) not visibly pregnant. We fitted individuals sites, and recorded location (GPS coordinates). with a single-stage transmitter on a replaceable cable tie weighing 2.5–3.5 g (Sirtrack Limited, Hawkes Bay, NZ) in the Assessment of shelter sites field before release at the point of capture. We tracked animals We assessed shelter sites for vegetation density, height and to their diurnal shelter site daily for a period of one month using structure. We measured vegetation density with a light meter a Yagi hand-held antenna and a communications specialist (in lux) at 20 cm intervals of a 1 m profile (in duplicate) through R1000 receiver (Sirtrack Limited). Transmission was highly the centre of the shelter site. We used an umbrella to shade the directional at close range allowing us to pinpoint animals to light meter from direct sunlight to enable assessment to occur in
Pereoglou, F., Population processes in an early successional heathland 146 Appendix 1:Manuscript - Refuge site selection
Refuge site selection by the eastern chestnut mouse Wildlife Research 293 diffuse light conditions on any given day or time of day. We also Shelter site selection recorded the maximum height of vegetation and the structure(s) We identified 453 unique shelter sites from 1300 tracking providing the greatest cover. We searched shelter sites for the fixes. On average, each animal had 8 Æ 0.5 shelter sites, spent presence of a mouse-made nest structure. For each shelter site, 2.5 Æ 0.1 days (total consecutive and non-consecutive fixes) we also assessed a control site using the identical procedure. We in a shelter site and re-visited shelter sites 0.8 Æ 0.1 times located control sites at a random compass bearing and a random (defined as the number of times a mouse returned to a distance between 2 and 20 m from a known shelter site. previously identified shelter site). Above ground nest structures were present in 15% of shelter sites. We did not locate any below ground burrow systems. Data analysis We found shelter sites had lower lux values (= more dense We used paired t-tests to determine whether shelter sites differed vegetation) up to 60 cm above the ground (Fig. 2) and taller from controls in vegetation density (estimated as mean lux at 0, vegetation with more structure compared with control sites 20, 40, 60, 80 and 100 cm), vegetation height and total structure(s) (Table 2). The type of structure present also differed between present. We used two-sample binomial tests (Collett 2002)to shelter and control sites with Xanthorrhea, buttongrass and dead determine whether mice preferentially selected the presence of stems preferentially selected by the eastern chestnut mouse structures of different types. To identify effects of sex and/or (Table 3). These structures were significant predictors of season in the selection of shelter sites, we used linear mixed decreasing standardised light intensity when we analysed models (McCulloch et al. 2008) with the response variables control site data separately (Table 4). being vegetation height and log-transformed standardised light intensity (as a surrogate for vegetation density, calculated as the proportion of light reaching the ground layer relative to what was 20.0 available at 1 m). We fitted animal identifier and site as random terms to evaluate individual animal and site variation. We 17.5 lux) excluded control sites from this analysis. 3 15.0 Using control sites only, we investigated vegetation structure 12.5 for its effect on standardised light intensity (where standardised light intensity approaching zero would approximate greater 10.0 ‘cover’). We fitted all possible regression models and ranked 7.5 them using the Schwarz information criterion (SIC) (Schwarz 1978). The candidate variables were the representative structural 5.0 categories: Xanthorrhea, buttongrass, dead stems, shrub and no Light intensity (N x 10 2.5 cover. We considered a primary refuge to be one with a nest 0.0 0 0.2 0.4 0.6 0.8 1.0 structure present. Mouse-made nest structures consisted of finely processed dried grass or sedge woven into either (i) a Profile height (m) pad ~10 cm in diameter, or (ii) a sphere ~10 cm in diameter and Fig. 2. Light intensity measured in lux at 20 cm intervals of a 1 m profile height with an inconspicuous opening ~2 cm wide. We fitted through eastern chestnut mouse shelter sites (solid line) and paired control generalised linear models (GLMs) (McCullagh and Nelder sites (broken line). Loss of light is a surrogate for vegetation density. Bars 1989) assuming a Bernoulli response with a logit link function indicate mean Æ s.e. Values were significantly different at the 0–0.6 m height for the probability of occurrence of a nest. The candidate categories. predictors were sex, season, habitat attributes (vegetation density = log-transformed standardised light intensity, Table 2. Means (þs.e.), t-statistics and significance levels for vegetation height and structure) and fidelity parameters (re- differences in height and total number of structures present between use, count of fixes standardised by number of radio- paired shelter and control sites tracking days). We ranked all possible models using SIC. Variate Shelter Control tP Height (m) 0.869 ± 0.012 0.785 ± 0.013 4.99 <0.001 Results Total structures 1.589 ± 0.028 1.243 ± 0.023 10.43 <0.001 Trapping and radio-tracking We attached radio-collars to 58 of the 140 individuals of Table 3. Proportion of successes, test statistic and significance level for eastern chestnut mouse that we captured. We summarised data differences in the types of structures present in shelter and control sites on the number of mice captured and radio-collared at each site in Structure type Shelter Control s.e. difference P each season (Table 1). On average, we tracked each animal 22 Æ 1.53 times (mean Æ s.e.) to diurnal shelter sites. The Xanthorrhea 0.496 0.211 0.029 <0.001 overall success rate of radio-tracking individuals for >20 Buttongrass 0.068 0.008 0.123 <0.001 fi Dead stems 0.380 0.257 0.030 <0.001 consecutive xes was 67% with the remaining animals tracked < less intensively due to predation, radio-collar loss or signal Shrub 0.384 0.696 0.031 0.001 No cover 0.255 0.458 0.030 <0.001 failure.
Pereoglou, F., Population processes in an early successional heathland 147 Appendix 1:Manuscript - Refuge site selection
294 Wildlife Research F. Pereoglou et al.
Table 4. Multiple regression statistics for a test of the effect of type of Table 5. Parameters for predicting the presence of a primary refuge. structure on standardised light intensity. Analysis excluded shelter Season refers to the difference between summer breeding and winter sites due to mouse-selection bias non-breeding seasons
Parameter Estimate s.e. t-statistic P Parameter Estimate s.e. Statistic P Constant –1.25 0.025 –49.98 <0.001 Constant –3.69 0.90 –4.09 <0.001 Xanthorrhea –0.39 0.047 –8.28 <0.001 Log10 sLI –0.634 0.39 –1.61 0.108 Buttongrass –1.08 0.205 –5.24 <0.001 STD count of fixes 0.74 1.25 0.59 0.553 Dead stems –0.13 0.044 –2.91 0.004 Season –3.22 1.42 –2.27 0.023 Log10 sLI * season –1.548 0.59 –2.60 0.009 STD count of fixes * season 3.58 1.86 1.93 0.054 Irrespective of individual animal and site variation, male and female mice did not differ in the standardised light intensity of sites they selected for sheltering but showed a construct a nest in a shelter site if it comprised very dense tendency to choose sites with lower standardised light vegetation (Table 5). During the breeding season, mice did not intensity during winter (F1,88 = 9.6, P = 0.003; Fig. 3). The appear to be as selective and any given shelter site was equally height of shelter sites did not differ significantly between sexes likely to have a nest structure. or across seasons. The eastern chestnut mouse was generally a solitary nester. We observed 1165 instances of solitary refuge use. We observed Refuge use and fidelity only very occasionally shelter site sharing and confirmed one record of cohabitation. The most common types of sharing we We were able to predict a primary refuge (i.e. those scored present recorded were: (i) where two different mice used the same for a nest structure) based on log-transformed standardised light shelter site during different seasons (mice ‘inherited’ the intensity and season (Table 5). We found primary refugia were shelter site from a previous occupant – 14 observations); or characterised by vegetation that was denser than secondary (ii) where two different mice used the same shelter site at shelters (i.e. mouse-selected shelter sites that did not have a different times during the same season of radio-tracking nest structure present). This relationship was significant in the (sequential shelter site use – 14 observations; six in summer non-breeding season indicating mice preferentially chose to and nine in winter). We observed a male–female pair of radio- collared mice in the same shelter site on one occasion for 14 –1.9 contiguous daily fixes during winter. This was the only record of
–1.95 cohabitation. We were unable to determine the genetic relationship (if any) between these two mice. We also –2.0 recorded seven observations of long-term shelter site fidelity where the same individual was radio-tracked in multiple –2.05 seasons and used the same shelter sites in both periods. This is –2.1 potentially more common than recorded here due to the small number of mice we re-collared and radio-tracked during both –2.15 light intensity (lux) seasons.
Log–scale standardised –2.2 Discussion –2.25 Refuge site selection
1.0 female Selection of appropriate refuge sites represents a substantial male investment of time and energy by an animal. Animals spend 0.95 periods of inactivity in these locations and shelter and raise their young in them. Therefore, the construction of nests and selection of refuge sites should afford the best protection from 0.9 predation for both adults and offspring to maximise fitness (Stokes 1995). The eastern chestnut mouse preferentially 0.85
Height (m) selected refuge sites with particular habitat attributes. We found shelter sites were selected that had taller, denser 0.8 vegetation compared with the surrounding habitat. Sites with multiple structures, such as Xanthorrhea, Gymnoschoenus 0.75 (buttongrass) and collapsed dead shrubs, were also often Summer Winter selected. We suggest these features have a combined effect in providing safety from trampling and predation, insulation against Season unfavourable climatic conditions, access to foraging areas and Fig. 3. Log-scale standardised light intensity (sLI) of male and protection from aggressive interactions with competitors. For female eastern chestnut mouse shelter sites (Æs.e.) selected during the eastern chestnut mouse, the selection of dense, protected different seasons. areas for shelter may reflect competitor avoidance. Previous
Pereoglou, F., Population processes in an early successional heathland 148 Appendix 1:Manuscript - Refuge site selection
Refuge site selection by the eastern chestnut mouse Wildlife Research 295 studies indicate competitive exclusion by the swamp rat Many mammals defend multiple, rarely occupied refuge (Rattus lutreolus) on eastern chestnut mouse access to food sites (Happold 1976; Cockburn 1981a; Wolff and Hurlbutt resources (Luo et al. 1994; Luo and Fox 1995) and preferred 1982; Lindenmayer et al. 1996; Hall and Morrison 1997; habitat (Higgs and Fox 1993; Morris et al. 2000). In our study, the Moro and Morris 2000; Woods and Ford 2000; Morzillo et al. swamp rat was absent from four of the six study sites. Refugia 2003; Sanecki et al. 2006; Hinkelman and Loeb 2007; selected by the eastern chestnut mouse in the two sites where Ebensperger et al. 2008). Refuge fidelity and site attachment swamp rats were present were not consistently different from has links to social organisation and animal behaviour. The those selected in the sites where the swamp rat was absent. From eastern chestnut mouse was transient in their refuge use. Our the results of our study, it is impossible to determine the extent observations indicate the eastern chestnut mouse was most of competition for shelter. However, dense cover has been commonly a solitary nester. Solitary sheltering might reflect shown to be important for the swamp rat (Catling 1986; sparse food resource availability or a territorial social Haering and Fox 1995; Monamy and Fox 2000; Fox and organisation (West and Dublin 1984; Ostfeld 1985). There Monamy 2007). The height of vegetation and areas of dense was some evidence for sharing of refuge sites and long-term cover have been shown to be an important factor in the choice of fidelity to particular sites. Long-term site fidelity is potentially refuge location for other terrestrial rodent species (Frank and more common than was possible to record in this study due to Layne 1992; Stokes 1995; Gray et al. 1998; Wagner et al. 2000). the small number of animals we re-collared during both seasons. The silky mouse’s(P. apodemoides) preference for low, dense The most commonly recorded instances of shelter site sharing vegetation and the availability of desert banksia (B. ornata) was were temporal segregation or inheritance of a shelter site from a considered a major influence on the distribution and location of previous occupant. There was only one record of a pair-bond and refugia (Cockburn 1981a). this was observed during winter. The benefits of refuge site fidelity and defence include: (i) highlighting ownership of, and defence of a shelter site against conspecifics; (ii) reduced fi Refuge site use and delity parasite burden; (iii) avoiding predators learning patterns of Daytime refuges can include both long-term shelter sites and emergence behaviour; and (iv) periodic change in foraging temporary resting places (Wolff and Hurlbutt 1982). Given the area utilised (Lindenmayer et al. 1996). Familiarity with an multiple functions of a refuge site, we expected to see a area and a known location of proven shelter is probably primary–secondary model of refuge use. Refuges offering the crucial to survival when exploring unknown territory (Gray best protection are likely to be used most frequently and for long et al. 1998). Shelter site replacement (occupation by alternate periods of time. The acquisition of refugia has a spatial element. animals) is not necessarily a random event (Haim and Rozenfeld To take advantage of dispersed food resources, animals might 1995). We propose that shelter sites used by different individuals have many, less frequently used refuge sites (Ward 1978). We of the eastern chestnut mouse in different seasons may result found evidence for hierarchical refuge selection. We considered from parent–offspring or sibling relationship between animals. the presence of a nesting structure to be indicative of a primary Bonds resulting from genealogical relationships refuge. The number of days spent in a shelter and whether a shelter (parent–offspring and sibling bonds) are not uncommon for site was re-visited during the course of the study were not pseudomyine rodents (Happold 1976). The silky mouse has significant predictors of a primary refuge. However, we did been recorded from communal burrows in groups comprising find a seasonal effect, with the density of the vegetation in a breeding pair with one or more litters of varying developmental winter being the best predictor of whether a shelter site would stages (Cockburn 1981b). The Hastings River mouse (P. oralis) be selected for building a nesting structure. Even though females has high shelter site fidelity (Townley 2000) and there is some produced litters during the breeding season, they were no more evidence to suggest offspring inherit the maternal range likely to build nesting structures than males. Any shelter site in (S. Townley pers. comm.). Further research is needed to the breeding season had an equal likelihood of being a primary confirm if this is the case in the eastern chestnut mouse and it refuge. We suggest this may be a result of (i) reproductive would be of interest to determine whether the relationship constraints including competition for mates, (ii) predator between animals is a significant determinant of spatial patterns avoidance during the summer months, (iii) competition for in sheltering. foraging resources among females weaning litters, and (iv) an insufficient survey of breeding females because we did not attach radio-collars to visibly pregnant animals. Retaining primary Prescribed burning practices refuge sites over winter is essential for animals to regulate Fire management in areas supporting populations of the eastern against thermal extremes and ensure access to foraging chestnut mouse should ensure the retention of tall, dense resources while competing with other adult animals, sub-adults vegetation and structural features in burnt landscapes. We and juvenile offspring from the previous breeding season. speculate that as time since fire increases, eastern chestnut Seasonality in shelter site use has been recorded for other mouse shelter sites have the potential to become a limiting rodent species (Madison et al. 1984; Wolton and Trowbridge resource. Vegetation regeneration in heathland would result in 1985; Bubela and Happold 1993) and was attributed to a sparse ground layer and in the complete absence of fire we environmental factors (i.e. shift in shelter site location could expect fewer structural components to be available. That is: dependent on ambient temperature (Madison et al. 1984)) and (i) the current collapsed shrub layer would decompose; functionality (i.e. different refuges for weaning litters (Morzillo (ii) although a long-lived, slow-growing species, the thatched, et al. 2003)) compared with refugia solely for sheltering. dead leaves of unburnt Xanthorrhea would no longer provide
Pereoglou, F., Population processes in an early successional heathland 149 Appendix 1:Manuscript - Refuge site selection
296 Wildlife Research F. Pereoglou et al. cover at ground level as plants grew taller (Gill and Ingwersen present at the ground layer for refuge. Refugia are highly 1976); and (iii) woody species would dominate or eliminate flammable and do not provide protection from fire. These herbaceous plants, increasing in biomass and height shelter site requirements have implications for land (Ingwersen 1977; Specht 1981; Keith et al. 2002). Conversely, management and prescribed burning practices. There is some immediately after a fire, it is also reasonable to expect shelter evidence for a primary–secondary model of refuge use. Primary sites and refugia to be limiting, particularly following a high refuge sites had a nesting structure present and were the more intensity burn. In contrast to other Pseudomys species (ash-grey dense shelter sites selected by the eastern chestnut mouse. Any mouse, P. albocinereus (Happold 1976), silky mouse (Cockburn shelter had an equal likelihood of being a primary or secondary 1981a), smoky mouse, P. fumeus (Woods and Ford 2000), New refuge during the summer breeding season but dense vegetation Holland mouse, P. novaehollandiae (Kemper 1981), and heath was the best predictor of a primary refuge for over-wintering rat, P. shortridgei (Taylor and Horner 1972; Happold 1976)) and animals. There are likely to be several reasons for this seasonal Fox (1995), we did not find the eastern chestnut mouse to use a aspect of shelter site use including reproductive constraints and below ground burrow complex. Burrows provide a more fire intra-specific competition for reproductive resources in the resilient shelter (Happold 1976; Kemper 1981; Friend 1993; summer breeding season and environmental constraints Sutherland and Dickman 1999; Whelan et al. 2002). With the including regulation against thermal extremes in the absence complete removal of above ground vegetation, the ability of of huddling behaviour in winter. The maintenance of multiple, refugial populations to survive in unburnt patches would be a rarely occupied shelters is consistent with data for other mammals critical factor affecting population persistence (Friend 1993; and is likely to be the result of territorial defence, access to Whelan 1995). In a manipulative experiment designed to dispersed resources and predator avoidance. The eastern promote small mammal responses to a simulated fire, there chestnut mouse is mostly solitary in its sheltering habits but was no effect of a reduction in cover on the eastern chestnut there is some evidence for short-term sharing, temporal mouse (Fox et al. 2003). However, the spatial scale and structural segregation in the use of the same sites, or familial inheritance effect of wildfire is likely to be different from that of experimental in the longer time frame. clipping, which, for instance, contained intact patches of vegetation within treatment plots (i.e. buttongrass). Further studies are necessary to determine whether the changes in Acknowledgements fi vegetation regrowth after re affect habitat preferences of We most gratefully acknowledge the Wreck Bay Aboriginal Community and shelter sites and refugia. the Wandandian People of the Dharawal and Dhurga language groups on Assuming the occurrence of fire in heathland is important for whose lands this study was conducted. For valuable field assistance, advice retention of habitat features selected by the eastern chestnut and discussions, we thank Wendy Hartman, Alfred Wellington, Tony Carter, mouse for refugia, appropriate timing of prescribed burning Geoff Kay, Steve Phillips, Marama Hopkins, Nick Dexter, Karen Viggers, practices may be important. Within mouse-selected refuge Sally Townley, Les Mitchell and Martin Westgate. This research was funded sites, those with denser vegetative cover were used for over- by the Norman Wettenhall Foundation and the MA Ingram Trust and wintering. These shelters were significantly more likely to have a conducted under The Australian National University Animal nest structure present. The eastern chestnut mouse builds an Experimentation Ethics Committee protocol C.RE.49.08, NSW National fi above ground surface nest comprising dried sedge or grassy Parks and Wildlife Service Scienti c Licence S12690 and Booderee National Park Permit BDR08/00009. material woven into a pad or spherical structure. These are highly flammable refuges. Protection from fire by such a shelter would be limited and the likelihood of direct mortality References by fire would be very high. The eastern chestnut mouse nest structure resembles that of the eastern bristle bird (Dasyornis Banks, S. C., Dujardin, M., McBurney, L., Blair, D., Barker, M., and brachypterus); a co-occurring heathland species, which has also Lindenmayer, D. B. (2011). Starting points for small mammal fi been found absent immediately after wildfire from sites that were population recovery after wild re: recolonisation or residual – previously occupied (Lindenmayer et al. 2009). The only other populations? Oikos 120(1), 26 37. doi:10.1111/j.1600-0706.2010. 18765.x Pseudomys species known exclusively to build an above ground Braithwaite, R. W. (1990). Australia’s unique biota: implications for grass nest is the western chestnut mouse (P. nanus), which is also ecological processes. Journal of Biogeography 17, 347–354. now extinct from the southern part of the species’ historical range doi:10.2307/2845364 and is rare or extinct throughout current distribution (Strahan Breed, B., and Ford, F. (2007). ‘Native Mice and Rats.’ (CSIRO Publishing: 1995; Breed and Ford 2007). Management practices that allow Melbourne.) burning during winter months would remove valuable primary Bubela, T. M., and Happold, D. C. D. (1993). The social organisation and refuge sites and may result in the post-fire residual eastern mating system of an Australian sub-alpine rodent, the broad-toothed rat chestnut mouse population expending greater energy to re- Mastacomys fuscus Thomas. Wildlife Research 20(4), 405–417. establish over-wintering nesting structures at a cost of reduced doi:10.1071/WR9930405 fi foraging capacity. Catling, P. C. (1986). Rattus lutreolus, coloniser of heathland after re in the absence of Pseudomys species. Australian Wildlife Research 13(2), – 127 139. doi:10.1071/WR9860127 Conclusions Catling, P. C. (1991). Ecological effects of prescribed burning practices on the mammals of southeastern Australia. In ‘Conservation of Australia’s The eastern chestnut mouse preferentially selects relatively Forest Fauna’. (Ed. D. Lunney.) pp. 353–363. (Royal Zoological Society taller, denser vegetation (up to 1 m) with structural features of NSW: NSW.)
Pereoglou, F., Population processes in an early successional heathland 150 Appendix 1:Manuscript - Refuge site selection
Refuge site selection by the eastern chestnut mouse Wildlife Research 297
Cockburn, A. (1981a). Diet and habitat preference of the silky desert mouse Hinkelman, T. M., and Loeb, S. C. (2007). Effect of woody debris abundance Pseudomys apodemoides (Rodentia). Australian Wildlife Research 8(3), on daytime refuge use by cotton mice. Southeastern Naturalist (Steuben, 475–497. doi:10.1071/WR9810475 ME) 6(3), 393–406. doi:10.1656/1528-7092(2007)6[393:EOWDAO] Cockburn, A. (1981b). Population processes of the silky desert mouse 2.0.CO;2 Pseudomys apodemoides (Rodentia) in mature heathlands. Australian Ingwersen, F. (1977). Vegetation development after fire in the Jervis Bay Wildlife Research 8(3), 499–514. doi:10.1071/WR9810499 Territory. Masters Thesis, The Australian National University, Canberra. Cockburn, A., Braithwaite, R. W., and Lee, A. K. (1981). The response of the Keith, D. A., McCaw, W. L., and Whelan, R. J. (2002). Fire regimes in heath rat, Pseudomys shortridgei, to pyric succession – a temporally Australian heathlands and their effects on plants and animals. In dynamic life-history strategy. Journal of Animal Ecology 50(3), ‘Flammable Australia: the Fire Regimes and Biodiversity of a 649–666. doi:10.2307/4127 Continent’. (Eds R. A. Bradstock, J. E. Williams and A. M. Gill.) Collett, D. (2002). ‘Modelling Binary Data.’ 2nd edn. (Chapman and Hall: pp. 199–237. (Cambridge University Press: New York.) London.) Kemper, C. M. (1981). Description of Pseudomys novaehollandiae burrows Ebensperger, L. A., Sobrero, R., Campos, V., and Giannoni, S. M. (2008). located with radioisotopes. Australian Mammalogy 4, 141–143. Activity, range areas, and nesting patterns in the viscacha rat Octomys Klein, H. G., and Layne, J. N. (1978). Nesting behaviour in four species of mimax. Journal of Arid Environments 72(7), 1174–1183. doi:10.1016/j. mice. Journal of Mammalogy 59(1), 103–108. doi:10.2307/1379879 jaridenv.2008.02.003 Letnic, M. (2003). The effects of experimental patch burning and rainfall Fox, B. J. (1982). Fire and mammalian secondary succession in an Australian on small mammals in the Simpson Desert, Queensland. Wildlife Research coastal heath. Ecology 63(5), 1332–1341. doi:10.2307/1938861 30(6), 547–563. doi:10.1071/WR02093 Fox,B. J.(1995).Eastern chestnut mousePseudomysgracilicaudatus.In‘The Letnic, M., Dickman, C. R., Tischler, M. K., Tamayo, B., and Beh, C. L. Mammals of Australia’. 2nd edn. (Ed. R. Strahan) pp. 601–3. (Reed: (2004). The responses of small mammals and lizards to post-fire Sydney.) succession and rainfall in arid Australia. Journal of Arid Environments Fox, B. J., and Monamy, V. (2007). A review of habitat selection by the 59(1), 85–114. doi:10.1016/j.jaridenv.2004.01.014 swamp rat Rattus lutreolus (Rodentia : Muridae). Austral Ecology 32(8), Lindenmayer, D. B., Welsh, A., Donnelly, C. F., and Meggs, R. A. (1996). 837–849. doi:10.1111/j.1442-9993.2007.01849.x Use of nest trees by the mountain brushtail possum (Trichosurus Fox, B. J., Taylor, J. E., and Thompson, P. T. (2003). Experimental caninus) (Phalangeridae : Marsupialia). I. Number of occupied trees manipulation of habitat structure: a retrogression of the small mammal and frequency of tree use. Wildlife Research 23(3), 343–361. doi:10.1071/ succession. Journal of Animal Ecology 72(6), 927–940. doi:10.1046/ WR9960343 j.1365-2656.2003.00765.x Lindenmayer, D. B., Cunningham,R. B., and Peakall, R. (2005). The recovery Frank, P. A., and Layne, J. N. (1992). Nests and daytime refugia of cotton of populations of bush rat Rattus fuscipes in forest fragments following mice (Peromyscus gossypinus) and golden mice (Ochrotomys nuttalli)in major population reduction. Journal of Applied Ecology 42(4), 649–658. south central Florida. American Midland Naturalist 127(1), 21–30. doi:10.1111/j.1365-2664.2005.01054.x doi:10.2307/2426318 Lindenmayer, D. B., MacGregor, C., Welsh, A., Donnelly, C., Crane, M., Friend, G. R. (1993). Impact of fire on small vertebrates in mallee woodlands Michael, D., Montague-Drake, R., Cunningham, R. B., Brown, D., and heathlands of temperate Australia: a review. Biological Conservation Fortescue, M., Dexter, N., Hudson, M., and Gill, A. M. (2008). 65(2), 99–114. doi:10.1016/0006-3207(93)90439-8 Contrasting mammal responses to vegetation type and fire. Wildlife Gill, A. M., and Ingwersen, F. (1976). Growth of Xanthorrhoea australis R. Research 35(5), 395–408. doi:10.1071/WR07156 Br. in relation to fire. Journal of Applied Ecology 13(1), 195–203. Lindenmayer, D. B., MacGregor, C., Wood, J. T., Cunningham, R. B., Crane, doi:10.2307/2401938 M., Michael, D., Montague-Drake, R., Brown, D., Fortescue, M., Dexter, Gray, S. J., Hurst, J. L., Stidworthy, R., Smith, J., Preston, R., and N., Hudson, M., and Gill, A. M. (2009). What factors influence rapid MacDougall, R. (1998). Microhabitat and spatial dispersion of the post-fire site re-occupancy? A case study of the endangered eastern grassland mouse (Mus spretus Lataste). Journal of Zoology 246, bristlebird in eastern Australia. International Journal of Wildland Fire 299–308. doi:10.1111/j.1469-7998.1998.tb00160.x 18(1), 84–95. doi:10.1071/WF07048 Haering, R., and Fox, B. J. (1995). Habitat utilisation patterns of Luo, J., and Fox, B. J. (1994). Diet of the eastern chestnut mouse (Pseudomys sympatric populations of Pseudomys gracilicaudatus and Rattus gracilicaudatus). II. Seasonal and successional patterns. Wildlife lutreolus in coastal heathland: a multivariate analysis. Australian Research 21(4), 419–431. doi:10.1071/WR9940419 Journal of Ecology 20(3), 427–441. doi:10.1111/j.1442-9993.1995. Luo, J., and Fox, B. J. (1995). Competitive effects of Rattus lutreolus tb00559.x presence on food resource use by Pseudomys gracilicaudatus. Haim, A., and Rozenfeld, F. M. (1995). Temporal segregation in Australian Journal of Ecology 20(4), 556–564. doi:10.1111/j.1442-9993. coexisting Acomys species: the possible role of nest site. Journal of 1995.tb00575.x Arid Environments 29(4), 505–509. doi:10.1016/S0140-1963(95) Luo, J., Fox, B. J., and Jefferys, E. (1994). Diet of the eastern chestnut 80022-0 mouse (Pseudomys gracilicaudatus). I. Composition, diversity and Hall, L. S., and Morrison, M. L. (1997). Den and relocation site characteristics individual variation. Wildlife Research 21(4), 401–417. doi:10.1071/ and home ranges of Peromyscus truei in the White Mountains of WR9940401 California. The Great Basin Naturalist 57(2), 124–130. Madison, D. M., Hill, J. P., and Gleason, P. E. (1984). Seasonality in the Happold, M. (1976). Social behavior of conilurine rodents (Muridae) of nesting behavior of Peromyscus leucopus. American Midland Naturalist Australia. Zeitschrift Fur Tierpsychologie – Journal of Comparative 112(1), 201–204. doi:10.2307/2425476 Ethology 40(2), 113–182. doi:10.1111/j.1439-0310.1976.tb00930.x Masters, P. (1993). The effects of fire driven succession and rainfall on Hartung, T. G., and Dewsbury, D. A. (1979). Nest-building behavior in seven small mammals in spinifex grassland at Uluru National Park, Northern species of muroid rodents. Behavioral and Neural Biology 27(4), Territory. Wildlife Research 20(6), 803–813. doi:10.1071/WR9930803 532–539. doi:10.1016/S0163-1047(79)92166-6 McCullagh, P., and Nelder, J. A. (1989). ‘Generalized Linear Models.’ 2nd Higgs, P., and Fox, B. J. (1993). Interspecific competition: a mechanism for edn. (Chapman and Hall: New York.) rodent succession after fire in wet heathland. Australian Journal of McCulloch, C. E., Searle, S. R., and Neuhaus, J. M. (2008). ‘Generalized, Ecology 18(2), 193–201. doi:10.1111/j.1442-9993.1993.tb00443.x Linear, and Mixed Models.’ 2nd edn. (Wiley: Hoboken, NJ, USA.)
Pereoglou, F., Population processes in an early successional heathland 151 Appendix 1:Manuscript - Refuge site selection
298 Wildlife Research F. Pereoglou et al.
Monamy, V., and Fox, B. J. (2000). Small mammal succession is determined Thompson, M. B., Medlin, G., Hutchinson, R., and West, N. (1989). Short- by vegetation density rather than time elapsed since disturbance. Austral term effects of fuel reduction burning on populations of small terrestrial Ecology 25(6), 580–587. doi:10.1111/j.1442-9993.2000.tb00063.x mammals. Australian Wildlife Research 16(2), 117–129. doi:10.1071/ Monamy, V., and Fox, B. J. (2010). Responses of two species of heathland WR9890117 rodents to habitat manipulation: vegetation density thresholds and the Townley, S. J. (2000). The ecology of the Hastings River mouse Pseudomys habitat accommodation model. Austral Ecology 35(3), 334–347. oralis (Rodentia : Muridae) in northeastern New South Wales and doi:10.1111/j.1442-9993.2009.02042.x southeastern Queensland. PhD Thesis, Southern Cross University, Moro, D., and Morris, K. (2000). Movements and refugia of Lakeland Downs Lismore. short-tailed mice, Leggadina lakedownensis, and house mice, Mus Townley, S. (2007). ‘Targeted Survey for the Eastern Chestnut Mouse domesticus, on Thevenard Island, Western Australia. Wildlife Research Pseudomys gracilicaudatus, New Holland Mouse Pseudomys 27(1), 11–20. doi:10.1071/WR99016 novaehollandiae, Eastern Pygmy-possum Cercartetus nanus, and Morris, D. W., Fox, B. J., Luo, J., and Monamy, V. (2000). Habitat-dependent Long-nosed Bandicoot Perameles nasuta on Pacific City Estate Lands, competition and the coexistence of Australian heathland rodents. Oikos Jervis Bay National Park.’ (NSW Department of Environment.) 91, 294–306. doi:10.1034/j.1600-0706.2000.910210.x Wagner, D. M., Feldhamer, G. A., and Newman, J. A. (2000). Microhabitat Morzillo, A. T., Feldhamer, G. A., and Nicholson, M. C. (2003). Home range selection by golden mice (Ochrotomys nuttalli) at arboreal nest sites. and nest use of the golden mouse (Ochrotomys nuttalli) in southern American Midland Naturalist 144(1), 220–225. doi:10.1674/0003-0031 Illinois. Journal of Mammalogy 84(2), 553–560. doi:10.1644/1545- (2000)144[0220:MSBGMO]2.0.CO;2 1542(2003)084<0553:HRANUO>2.0.CO;2 Ward, G. D. (1978). Habitat use and home range of radio-tagged opossums Newsome, A. E., and Catling, P. C. (1983). Animal demography in relation to Trichosurus vulpecula (Kerr) in New Zealand lowland forest. In ‘The fire and shortage of food: some indicative models. Ecological Studies 43, Ecology of Arboreal Folivores’. (Ed. G. G. Montogomery.) pp. 305–313. 490–505. (Smithsonian Institute Press: Washington DC.) Newsome, A. E., McIlroy, J., and Catling, P. C. (1975). The effects of an West, S. D., and Dublin, H. T. (1984). Behavioural strategies of small extensive wildfire on populations of twenty ground vertebrates in mammals under winter conditions: solitary or social? In ‘Proceedings southeast Australia. Proceedings of the Ecological Society of Australia of the International Colloquium: Winter Ecology of Small Mammals’. 9, 107–123. (Ed. J. F. Merritt.) pp. 293–299. (Carnegie Museum of Natural History: Ostfeld, R. S. (1985). Limiting resources and territoriality in microtine Pittsburgh, PA, USA.) rodents. American Naturalist 126(1), 1–15. doi:10.1086/284391 Whelan, R. J. (1995). ‘The Ecology of Fire.’ (Cambridge University Press: Sanecki, G. M., Green, K., Wood, H., Lindenmayer, D., and Sanecki, K. L. Cambridge.) (2006). The influence of snow cover on home range and activity of the Whelan, R. J., Rodgerson, L., Dickman, C. R., and Sutherland, E. F. (2002). bush-rat (Rattus fuscipes) and the dusky antechinus (Antechinus Critical life cycles of plants and animals: developing a process-based swainsonii). Wildlife Research 33(6), 489–496. doi:10.1071/WR05012 understanding of population changes in fire-prone landscapes. In Schwarz, G. (1978). Estimating the dimension of a model. Annals of Statistics ‘Flammable Australia: the Fire Regimes and Biodiversity of a 6, 461–464. doi:10.1214/aos/1176344136 Continent’. (Eds R. A. Bradstock, J. E. Williams and A. M. Gill.) Specht, R. L. (1981). Responsesto fire in heathlands and related shrublands. In pp. 94–124. (Cambridge University Press: New York.) ‘Fire and the Australian Biota’. (Eds A. M. Gill, R. H. Groves and Wolfe, J. L. (1970). Experiments on nest-building behaviour in Peromyscus I. R. Noble.) pp. 395–415. (Australian Academy of Science: Canberra.) (Rodentia : Cricetinae). Animal Behaviour 18(Part 4), 613–615. Stokes, M. K. (1995). Selection of refuge sites by sympatric Microtus doi:10.1016/0003-3472(70)90002-3 ochrogaster and Sigmodon hispidus. Journal of Mammalogy 76(1), 83. Wolff, J. O., and Hurlbutt,B. (1982).Day refugesof Peromyscus leucopusand doi:10.2307/1382317 Peromyscus maniculatus. Journal of Mammalogy 63(4), 666–668. Strahan, R. (Ed.) (1995). ‘The Mammals of Australia (2nd edn).’ (Reed: doi:10.2307/1380276 Chatswood.) Wolton, R. J., and Trowbridge, B. J. (1985). The ranging and nesting behavior Sutherland, E. F., and Dickman, C. R. (1999). Mechanisms of recovery after of wood mice, Apodemus sylvaticus (Rodentia : Muridae), as revealed by fire by rodents in the Australian environment: a review. Wildlife Research radio-tracking. Journal of Zoology 206, 203–224. 26(4), 405–419. doi:10.1071/WR97045 Woods, R. E., and Ford, F. D. (2000). Observations on the behaviour of the Taws, N. (1997). Vegetation survey and mapping of Jervis Bay Territory. smoky mouse Pseudomys fumeus (Rodentia: Muridae). Australian Environment Australia Report, Canberra. Mammalogy 22,35–42. Taylor, J. M., and Horner, B. E. (1972). Observations on reproductive biology of Pseudomys (Rodentia : Muridae). Journal of Mammalogy 53(2), 318–328. doi:10.2307/1379168 Thompson,P.T., andFox, B.J. (1993).Asymmetric competitionin Australian heathland rodents: a reciprocal removal experiment demonstrating the influence of size-class structure. Oikos 67(2), 264–278. doi:10.2307/ 3545471 Manuscript received 14 January 2011, accepted 28 May 2011
http://www.publish.csiro.au/journals/wr
Pereoglou, F., Population processes in an early successional heathland 152
Appendix 2: Manuscript – Site occupancy
5
Manuscript: Site occupancy of an early successional specialist Published, July 2015
10
Pereoglou, F., Population processes in an early successional heathland 153 Author's personal copy Appendix 2: Manuscript – Site occupancy
Landscape Ecol (2016) 31:255–269 DOI 10.1007/s10980-015-0240-2
RESEARCH ARTICLE
Landscape, fire and habitat: which features of recently burned heathland influence site occupancy of an early successional specialist?
Felicia Pereoglou . Christopher MacGregor . Sam C. Banks . Jeff Wood . Fred Ford . David B. Lindenmayer
Received: 10 October 2013 / Accepted: 14 July 2015 / Published online: 21 July 2015 Ó Springer Science+Business Media Dordrecht 2015
Abstract to occupancy. We also explored predictability in Context Multiple ecological drivers generate spatial patterns of small mammal species co-occurrence. patterns in species’ distributions. Changes to natural Results Landscape structure (patch area, landscape disturbance regimes can place early successional heterogeneity) was strongly related to site occupancy. habitat specialists at an increased risk of extinction Site occupancy was associated with dead shrubs in the by altering landscape patterns of habitat suitability. understory and rock cover on ground layer, but was not Objectives We developed a series of hypotheses to directly influenced by recent or historical fire. Con- evaluate the effects of landscape structure, fire history, trary to contemporary ecological theory, we found no and site-level habitat quality on site occupancy by an predictable species associations in our early succes- early successional specialist, the eastern chestnut sional community. mouse (Pseudomys gracilicaudatus). Conclusions We recommend surveys take account Methods We obtained eight years of monitoring data of landscape configuration and proximity to suitable from 26 sites in recently burned heathland in southeast habitat for optimal results. Fire regimes expected to Australia. We used generalised linear models to promote eastern chestnut mouse population growth determine which explanatory variables were related should encourage the retention of critical habitat features rather than be based on temporal rates of successional stages. For management to adequately account for post-disturbance patterns in early succes- sional communities, a species-by-species, multi- Electronic supplementary material The online version of scaled approach to research is necessary. this article (doi:10.1007/s10980-015-0240-2) contains supple- mentary material, which is available to authorized users. Keywords Fire Á Succession Á Heathland Á Patch F. Pereoglou (&) Á C. MacGregor Á S. C. Banks Á area Á Landscape heterogeneity Á Habitat specialist J. Wood Á D. B. Lindenmayer Fenner School of Environment and Society, The Australian National University, Frank Fenner Building 141, Canberra, ACT 0200, Australia Introduction e-mail: [email protected] Early successional communities are the starting point F. Ford for recovery after major disturbance. These commu- Directorate of Heritage and Biodiversity Conservation, Defence Support Group, Brindabella Business Park, nities are comprised of residual individuals, colonising P.O. Box 7925, Canberra, ACT 2610, Australia opportunists and habitat specialist species (Connell 15 123
Pereoglou, F., Population processes in an early successional heathland 154 Author's personal copy Appendix 2: Manuscript – Site occupancy 256 Landscape Ecol (2016) 31:255–269 and Slatyer 1977; Swanson et al. 2011). Understand- 1982) to low (e.g. desert mouse (P. desertor), Letnic ing the ecological processes operating in early and Dickman 2005; brown antechinus (Antechinus successional communities is important, given these stuartii), bush rat (Rattus fuscipes), swamp rat (R. ephemeral habitats are confined to the variable length lutreolus), Lindenmayer et al. 2008a; silky mouse (P. of time between a disturbance, and re-establishment of apodemoides), Di Stefano et al. 2011). Changes to fire the later successional stages of community composi- regimes can reduce habitat suitability for some species tion (Swanson et al. 2011). Estimating the duration of (e.g. Fox 1982; Baker et al. 2010). the early successional community is difficult as it is Contingent upon a species’ location along the related to the type of ecosystem and is likely to continuum, increased fire frequency may improve fluctuate within ecosystems of the same type (Brad- population viability. For early successional species, stock et al. 2012); for example, years for heathlands fire suppression can lead to population extinction (Keith et al. 2002) compared to decades or even through habitat degradation and impaired demo- centuries in some forests (McCarthy et al. 1999). This graphic processes (Hutto 2008; Templeton et al. time frame is likely to vary geographically (Velle and 2011). For example, with the re-introduction of fire, Vandvik 2014), and be dependent on the type and Templeton et al. (2011) showed isolated patches could frequency of disturbance events (e.g. Sousa 1979; transition into a stable metapopulation with concomi- Delmoral and Bliss 1993; Rossi et al. 2007; Gill and tant changes in the amount and patterns of animal Allan 2008). movements across the landscape. Furthermore, fire- Heathland communities often occur in the early specialist fauna have been shown to occur with successional stage of regeneration after disturbance. In increasing likelihood as spatial proximity to fire, or Australia, heathlands are one of the most fire-prone fire severity increases (Hutto 2008), and have positive and flammable ecosystems (Williams 1995; Keith population growth rates in habitat created by fire (Rota et al. 2002). Following disturbance, these heathlands et al. 2014). For early successional heathland special- can accumulate sufficient plant biomass for fire ists, the local physical conditions that are associated ignition every 2 years (Williams 1995) and reach a with their presence are likely to be in a continual state mature to senescent phase after 25–50 years (Specht of change. The ability to reverse population declines 1981). The compositional dynamics of heathland are for these species is complicated by the unpredictable, primarily driven by the fire regime. High fire fre- spatially and temporally variable nature of their quency (i.e. every 2 years) can reduce a heathland to a habitats. However, whether fire plays a positive role sedgeland by eliminating woody shrubs and allowing in species occurrence and assists in maintaining herbaceous species to dominate (Williams 1995). habitat suitability in early successional communities Long intervals between fires (i.e. greater than is poorly understood. 20 years) will produce closed scrubs with diminished Research across multiple spatial scales is necessary understory floristic richness (Williams 1995; Keith for identifying the processes underlying species’ et al. 2002). Heathlands generally occur as patches of responses in patchy and temporally-dynamic commu- various sizes within a heterogeneous landscape (Wil- nities (Levin 1992; Thornton et al. 2011). Most liams 1995). theoretical and empirical research in disturbance Fire occurs over multiple spatial and temporal ecology has focussed on site-level disturbance history scales, with variable impacts on biodiversity and key (e.g. time since fire). However, multiple ecological ecosystem processes that are strongly site- and drivers together generate spatial patterns in species’ species-specific (Pickett and White 1985; Turner distributions (Turner 2005). For instance, research in et al. 1998; Whelan et al. 2002). For small mammals, landscape ecology has a strong focus on landscape patterns of population responses to fire have been pattern variables such as patch size and landscape studied over multiple decades (Newsome et al. 1975; heterogeneity, as well as site-level habitat quality Masters 1993; Letnic et al. 2013). Small mammals are (Burel and Baudry 2003). Here, we tested the effects associated with a range of fire frequencies; from high of landscape patterns on spatial processes by integrat- (e.g. heath rat (Pseudomys shortridgei), Cockburn ing fire history [sensu landscape legacy (Turner 2005) et al. 1981; eastern chestnut mouse (P. gracilicauda- or ecological memory (Peterson 2002)], with classic tus), New Holland mouse (P. novahollandiae), Fox landscape ecology themes of patch size and landscape
123
Pereoglou, F., Population processes in an early successional heathland 155 Author's personal copy Appendix 2: Manuscript – Site occupancy
Landscape Ecol (2016) 31:255–269 257 heterogeneity, as well as community ecology themes Vieira and Briani 2013), and breeding resource of interspecific interactions. We investigated land- requirements (Banks et al. 2005b; Watson et al. scape patterns of habitat suitability across multiple 2012). The floristic richness of early successional spatial scales using a species described as an early heathland provides a varied diet for an opportunistic successional specialist, the eastern chestnut mouse generalist herbivore like the eastern chestnut mouse (Pseudomys gracilicaudatus Gould 1845). Specifi- (Luo et al. 1994). Dense vegetation on the ground cally, we assessed a series of hypotheses relating to the layer and a collapsed dead shrub layer are significant perceived importance of fire history, landscape, com- components of the species’ diurnal shelter require- munity and site-level habitat features influencing site ments (Pereoglou et al. 2011). occupancy in an early successional heathland (Fig. 1). Hypothesis 3 (interspecific interactions) We pre- Hypothesis 1 (fire history) We predicted occu- dicted eastern chestnut mouse site occupancy was pancy was positively related to the distribution of the negatively related to swamp rat (Rattus lutreolus) most recent fire, positively related to a high incidence occurrence, but was not influenced by the occurrence of fires, and negatively related to the years since last of the bush rat (R. fuscipes) or brown antechinus fire (Fig. 1). Recent fire creates suitable habitat for (Antechinus stuartii) (Fig. 1). After a disturbance, early successional specialists (Monamy and Fox species are added to the community assemblage when 2000). Therefore, with either a recent fire or an their specific habitat requirements are met (habitat increased incidence of fire at a site, post-fire recruit- accommodation model; Fox 1982). Species replace- ment by residual survivors should offset the potential ment is a result of subsequent competitive interactions for local extinction. that occur with time since fire or vegetation succession (Fox et al. 2003). The swamp rat is the only one of the Hypothesis 2 (site-level habitat quality) We pre- co-occurring species that is similar to the eastern dicted occupancy was positively related to understory chestnut mouse in its resource requirements and has floristic richness, the number of dead shrubs in the also been shown to be competitively superior (Higgs understory, and live vegetative cover on the ground and Fox 1993). layer (Fig. 1). The distribution of species after distur- bance is dependent on the availability of suitable local Hypothesis 4 (landscape context) We predicted habitats, and has been linked to shelter (Fox et al. site occupancy was positively related to heathland 2003; Robinson et al. 2013), food (Luo and Fox 1996; patch area and negatively influenced by landscape
Fig. 1 Conceptual diagram showing the predicted effects of explanatory variables on eastern chestnut mouse site occupancy; black arrows indicate a positive effect, white arrows indicate a negative effect, grey arrows indicate no effect
20
123
Pereoglou, F., Population processes in an early successional heathland 156 Author's personal copy Appendix 2: Manuscript – Site occupancy 258 Landscape Ecol (2016) 31:255–269
heterogeneity (Fig. 1). The probability of species throughout much of its range. The species inhabits occurrence in a habitat patch varies as a function of heathland in southern Australia and has been recorded patch size (MacArthur and Wilson 1967) with a in woodland in the northern parts of its range (Fox species’ area sensitivity being greater when the 2008). Usually reported at low density, populations intervening matrix is inhospitable (Prugh et al. peak in abundance post-fire before declining as 2008). The eastern chestnut mouse is considered a vegetation regenerates (Fox 1982). Threats to the heathland habitat specialist (Fox 1982) but there has viability of populations of the species are thought to been no study of how habitat availability and hetero- include fire suppression, loss of heath habitat, and geneity influences the species’ local distribution predation (Fox 2008). A summarised description of across a landscape. the eastern chestnut mouse and the co-occurring small mammal species with which it is expected to have competitive interactions is provided (Table 1; a full Materials and methods version of species’ descriptions available in Appendix S1). Study species Study area and survey design The eastern chestnut mouse is an endemic Australian murid rodent. It has a disjunct distribution along the Booderee National Park is a *6500 ha reserve east coast of Australia (inset Fig. 2) and is absent located in the Jervis Bay region, approximately
Fig. 2 Map of Booderee national park indicating study sites and vegetation types. Inset shows the distribution of the eastern chestnut mouse and the study area location 123
Pereoglou, F., Population processes in an early successional heathland 157 Author's personal copy Appendix 2: Manuscript – Site occupancy
Landscape Ecol (2016) 31:255–269 259
200 km south of Sydney, Australia (Fig. 2). Booderee
), National Park is characterised by natural heterogene- (2008a) , Fox , Breed ity in vegetation with six broad vegetation types , , Cockburn (1992 1994) , Lunney (1987)
), Dickman (forest, rainforest, woodland, heathland, shrubland , Crowther and ), Lunney (1981) (1984)
, Watts and Aslin and sedgeland) identified in previous vegetation (2008 (1982)
(2002) surveys (Taws 1997) (Fig. 2). (2007 , Fox and Kemper (1971)
, Luo et al. ( , Robinson Monitoring heathland sites (n = 26) began in 2003
1982) as part of a long-term study on the effects of fire on 2008) 2008b) and Lazenby-Cohen Crowther Braithwaite ( and Ford Braithwaite ( (1982) (2008) (1981) Fox and Archer Watts and Aslin Fox ( Warneke Source vertebrate biota (see Lindenmayer et al. 2008a). Site monitoring was conducted annually during the sum- mer months for the period 2003–2011. Sites com- prised a single 100 m transect with: (i) a black plastic drift fence connecting 380 mm deep 9 300 mm diameter pitfall traps at 0 and 20 m, 40 and 60 m, communal nesting terrestrial, territorial territorial mostly territorial and 80 and 100 m; (ii) large wire cage traps (30 9 30 9 60 cm) placed at 0 m and 100 m with four small wire cage traps (20 9 20 9 50 cm) at 20 m intervals between; and (iii) ten aluminium box traps (Elliott Scientific Equipment, Australia) placed Insectivorous Terrestrial or arboreal, Diet Habit every 10 m along the transect line from 0 to 90 m (Fig. 3). We baited wire cage traps and aluminium box
(female) traps with rolled oats and peanut butter. Effort varied 2 Herbivorous Nocturnal, terrestrial, 1–21 Herbivorous (male) Partially1–2 diurnal, 1–2 Omnivorous Nocturnal, terrestrial, Lifespan (years) \ Breed in season of birth 3–5 Yes 8–10 No 1–5 (3) No 1–7 (5) Yes Litter size A, S S S A, S Breed: all year (A) or season (S) 200 200 35 (male) 20 (female) 90 (male) 75 (female) weight (grams) [ Forest Heathland Sedgeland Heathland Widespread Heathland Woodland ) ) ) Summarized description of the eastern chestnut mouse and co-occurring small mammal species )
Fig. 3 Layout of a site transect and the location of vegetation 25 stuartii mouse gracilicaudatus Antechinus R. lutreolus Rattus fuscipes Pseudomys Brown antechinus ( Swamp rat ( Bush rat ( Eastern chestnut ( Table 1 Species Habitat Average adult plots. Diagram is not to scale
123
Pereoglou, F., Population processes in an early successional heathland 158 Author's personal copy Appendix 2: Manuscript – Site occupancy
260 Landscape Ecol (2016) 31:255–269
Table 2 Description of explanatory variables used to predict eastern chestnut mouse site occupancy Variable Description Mean SE Min Max
Survey year 2003–2010/11 (n = 9) Site Unique site identifier (n = 26) Effort Total number of traps open 9 total number of nights traps were 63.30 2.09 0 138 open Species associations Eastern chestnut mouse Eastern chestnut mouse captures (new animals ? recaptures) 0.63 0.11 0 12 Bush rat Bush rat captures (new animals ? recaptures) 2.78 0.28 0 24 Swamp rat Swamp rat captures (new animals ? recaptures) 0.45 0.10 0 12 Brown antechinus Brown antechinus captures (new animals ? recaptures) 1.89 0.16 0 11 Fire history Burned 2003 Site burned (1) or not (0) in the December 2003 fire Number of previous fires Count of previous fires recorded at each site 4.54 0.12 1 9 Years since fire Number of years elapsed since the last fire at each site Site-level habitat quality Dead shrub Count of dead shrubsa 28.10 3.25 0 270 Floristic richness Number of understory speciesa 13.09 0.46 4.5 26.50 Percent cover Percent live cover of all understory species combineda 85.44 1.79 15 100 Percent rock Percent rock cover in ground layerb 1.24 0.39 0 24.38 Percent bare Percent cover of bare ground in ground layerb 17.41 2.12 0 81.88 Percent regenerating Percent cover of regenerating vegetation in ground layerb 16.71 2.63 0 100 vegetation Landscape structure Patch area Number of hectares of contiguous patch of vegetation of type i 88.59 8.74 0.93 362.20 Landscape heterogeneity Number of vegetation types in 500 m radius of site 4.00 0.07 2 6 Summary statistics calculated for data pooled over sites and survey years a Average of data measured in two 20 9 20 m plots b Average of data measured in four 1 9 1 m subplots between sites and survey years, but on average, we of all plant species combined (cover, Table 2). Values opened traps for three consecutive nights each survey were averaged across the two vegetation plots. Percent year. We checked all traps daily during the trapping rock cover, bare ground and regenerating shrubs in the period and recorded species, sex, age class, trap ground layer were visually estimated in four 1 9 1m number and trap type. Animals were marked to sub-plots nested within the larger vegetation plot, identify recaptures within a single survey year trap- values were averaged (Table 2). ping period before release at the point of capture. Fire history data was extracted from GIS layers Vegetation was sampled during the survey years of available for Booderee National Park using ArcGIS 2004/05, 2006/07, 2007/08, and 2009/10 in two 20 9 version 9.2 (ESRI 2009). At each site, for each survey 20 m plots located at the 20–40 m and 60–80 m year, we calculated the number of years elapsed since sections of each transect (Fig. 3). We counted the the last fire (years since fire, Table 2), and counted the number of dead shrubs (where a shrub was defined as number of fires affecting the site since fire mapping single woody plant either multi-stemmed at the base began in the mid-1950s (previous fires, Table 2). Sites or, if single stemmed, \2 m tall; dead shrubs, were scored with a 0 or 1 to indicate whether they were Table 2), the number of understory plant species burned in the most recent major fire in December 2003 present (floristic richness, Table 2), and visually (burned 2003, Table 2). 80 % of heathland sites were estimated the understory (\2 m) percent live cover burned in the 2003 fire. 123
Pereoglou, F., Population processes in an early successional heathland 159 Author's personal copy Appendix 2: Manuscript – Site occupancy
Landscape Ecol (2016) 31:255–269 261
Landscape structure was measured from GIS layers via other explanatory variables by testing whether the of the vegetation mapping of Taws (1997) available apparent effect of fire was removed when another for Booderee National Park, using ArcGIS version 9.2 explanatory variable was included in our models, and (ESRI 2009). As a measure of landscape heterogene- in turn fitting a model to test whether that variable was ity, we determined the number of vegetation types significantly affected by fire. Based on the explanatory within a polygon of 500 m radius around each site variables identified in the GLM of all subsets and the (landscape heterogeneity, Table 2). A 500 m radius univariate HGLMs, we then fitted a multivariate was chosen because the closest sites were 1 km apart. HGLM using the predictors from the ‘best’ model. We calculated the area of the mapped polygon of Finally, we fitted the captures of other small heathland within which a site was located (patch area, mammals (bush rat, swamp rat and brown antechinus) Table 2). Landscape heterogeneity and patch area as dependent variables in a series of separate models to were fixed for the study period. determine whether the eastern chestnut mouse responded differently compared to sympatric species Statistical analyses in relation to fire history, landscape structure and site- level habitat quality. We used Spearman’s rank correlation (Conover 1999) to identify redundant variables. We explored the individual significance of the Results explanatory variables (Table 2; Fig. 1) using a series of univariate hierarchical generalised linear models We completed a total of 14,812 trap nights and made (HGLMs; Lee et al. 2006) with an over-dispersed 1346 captures (including 366 recaptures) of the Poisson distribution and logarithm-link function. Our eastern chestnut mouse and co-occurring terrestrial model response variable, site occupancy, was the total small mammal species. number of eastern chestnut mouse captures across trap nights at each site. We fitted each explanatory variable Correlation between explanatory variables individually in separate models that included site as a random term to account for variation among sites, and We found a low level of correlation between explana- survey year as a fixed effect (except where site tory variables (Table 3). Variable pairs with the occupancy was predicted by years since the last fire, highest levels of correlation included patch area and survey year was fitted as a random term). We landscape heterogeneity, previous fires and landscape calculated effort as the total number of traps open at heterogeneity, years since last fire and burned 2003, each site multiplied by the total number of nights that and bush rat and effort. The landscape structure traps were open within a survey year and we included variables represented different effects and were not so the log of effort as an offset variable to account for highly correlated that inclusion together substantially unequal survey effort between sites and survey years reduced the reliability of the further modelling. The (Table S1). extent of the 2003 fire meant insufficient fire ages were To check for possible interrelationships between available to model a time since fire effect on the effects of the explanatory variables, we also fitted occupancy so we removed years since fire from generalized linear models (GLMs; McCullagh and further analysis and retained the binary variable. Nelder 1989) for all possible subsets of the set of potential explanatory variables (Table 1; Fig. 1). We Fire history evaluated the fits using Schwarz information criterion (SIC; Schwarz 1978) and Akaike’s information crite- Although the eastern chestnut mouse was only rion (AIC; Burnham and Anderson 2002). Complex observed on burned sites, the direct effect of burning fixed effects models are computationally much easier in 2003 was not significant. This partly reflects the to fit using GLMs. Terms that did not contribute small number of heathland sites that were not burned anything useful in the HGLMs might be retained, but it in the most recent fire. Whether a site was burned in was very unlikely we would omit a term that was 2003 had a positive effect on eastern chestnut mouse important. We also looked for an indirect effect of fire site occupancy when the percent rock cover on the 30 123
Pereoglou, F., Population processes in an early successional heathland 160 Author's personal copy Appendix 2: Manuscript – Site occupancy
262 Landscape Ecol (2016) 31:255–269 1 0.67 2 0.32 - 0.13 - 0.12 0.21 1 - 0.22 0.08 0.34 0.16 1 0.47 1 - - - 0.27 0.03 - - 0.07 0.13 1 0.09 1 - - 0.02 0.13 0.19 0.29 0.230.13 0.13 0.03 0.32 0.08 0.04 0.11 0.40 0.49 1 0.14 1 0.12 0.05 ------0.52 0.16 1 0.06 0.02 0.32 0.06 - 2 - 0.55 2 0.40 1 0.02 0.100.09 0.22 0.19 0.03 0.10 0.45 0.10 0.210.18 0.15 0.05 0.050.34 0.25 0.08 0.11 0.11 0.08 ------0.29 0.02 0.08 0.18 1 0.04 0.14 0.19 0.13 0.18 0.09 ------0.50 0.12 0.15 0.12 0.18 0.01 0.08 0.17 0.090.16 0.08 0.11 ------0.18 1 0.06 0.00 1 0.05 0.48 0.01 0.21 0.03 0.03 0.06 0.04 0.17 0.12 ------0.10 1 0.02 0.030.02 0.17 0.190.08 0.02 0.25 1 0.05 0.06 0.01 0.06 0.11 0.02 0.19 1 - 0.59 0.11 - - - 0.17 0.07 0.39 - 0.04- 0.19 0.15- 0.17 - 12345678910111213141516 0.5 are bold [ ± Correlation matrix showing Spearman’s rank correlation coefficients for explanatory variables Variable
12 Effort 3 Eastern chestnut mouse Bush rat 4 Swamp rat 56 Brown antechinus 7 Burned 2003 8 Number of previous fires Years since fire 9 Dead shrub 12 Percent bare 14 Percent regenerating15 vegetation Patch area 16 Landscape heterogeneity 10 Floristic richness 11 Percent cover 13 Percent rock Table 3 Correlated variables
123
Pereoglou, F., Population processes in an early successional heathland 161 Author's personal copy Appendix 2: Manuscript – Site occupancy
Landscape Ecol (2016) 31:255–269 263
Table 4 Ten best DSIC DAIC Model terms generalized linear models from fitting all possible 0 1.22 Survey year ? percent rock ? burned 2003 subsets of explanatory 1.37 0 Dead shrub ? survey year ? percent rock ? burned 2003 variables 2.44 1.07 Survey year ? percent regenerating vegetation ? percent rock ? burned 2003 3.17 1.80 Survey year ? percent rock ? percent bare ? burned 2003 3.84 2.47 Survey year ? percent cover ? percent rock ? burned 2003 4.09 2.72 Survey year ? floristic richness ? percent rock ? burned 2003 4.16 2.79 Survey year ? percent rock ? brown antechinus ? burned 2003 DSIC and DAIC are the differences in SIC and AIC 4.40 3.03 Survey year ? percent rock ? patch area ? burned 2003 for the given models from 4.41 3.04 Survey year ? percent rock ? bush rat ? burned 2003 the value for the optimal 4.45 3.08 Survey year ? percent rock ? burned 2003 ? landscape heterogeneity model
Fig. 4 Conceptual diagram showing the modelled effects of explanatory variables on eastern chestnut mouse site occupancy; black arrows indicate a positive effect, white arrows indicate a negative effect, broken line represents indirect effect
ground layer was included in the model (Table 4). number of dead shrubs in the understorey was However, all of the sites that were not burned in 2003 associated with a small increase in eastern chestnut had no rock, so the effects are confounded to some mouse captures per 100 trap nights (Fig. 5a). As an extent and it is mainly a percent rock cover effect. indirect effect of fire, we found the number of dead
Hence, H1 was rejected, as we were unable to detect a shrubs significantly increased with burning in 2003. direct effect of either previous fires or burning in the Percent rock cover on the ground layer positively 2003 fire on eastern chestnut mouse site occupancy influenced eastern chestnut mouse site occupancy in (Table S2; Fig. 4). burned sites (Table 4). Understory floristic richness and percent cover did not affect eastern chestnut Site-level habitat quality mouse site occupancy, nor did the percent bare ground or regenerating shrubs in the ground layer. We
The number of dead shrubs in the understory had a accepted H2, habitat features of the understory and weak positive effect on eastern chestnut mouse site ground layer influence eastern chestnut mouse site occupancy (Fig. 4). An exponential increase in the occupancy (Table S2; Fig. 4). 35
123
Pereoglou, F., Population processes in an early successional heathland 162 Author's personal copy Appendix 2: Manuscript – Site occupancy
264 Landscape Ecol (2016) 31:255–269
Table 5 Multivariate HGLM of significant explanatory variables Effect Estimate SE T P
Bush rat -0.116 0.051 -2.26 0.024 Burned 2003 1.035 0.653 1.59 0.113 Percent rock 0.022 0.031 0.72 0.471 Dead shrub 0.010 0.005 2.15 0.032 Patch area 0.004 0.001 2.86 0.004 Landscape heterogeneity -0.354 0.179 -1.98 0.048 Model included eastern chestnut mouse site occupancy as response variable, site as a random effect, survey year as a fixed effect, and the log of effort as an offset variable
(Table S2). Hence, H3 was rejected, as co-occurring small mammal species did not influence eastern chestnut mouse site occupancy as predicted (Table S2; Fig. 4). We found direct effects of fire history and land- scape structure for the sympatric small mammal species. Burning in the 2003 fire had a strong negative effect on both swamp rat and brown antechinus site occupancy estimates (Table S2). A weak negative effect of patch area on brown antechinus and swamp rat site occupancy indicates a potential sensitivity to landscape heterogeneity (Table S2).
Landscape structure
Landscape heterogeneity had a strong negative effect on eastern chestnut mouse site occupancy (Fig. 4), with the Fig. 5 Number of eastern chestnut mouse captures per 100 trap nights predicted from hierarchical generalised linear models number of captures per 100 trap nights decreasing over a range of values of significant explanatory variables; significantly as habitat diversity increased (Fig. 5b). To a number of dead shrubs, b patch area, and c landscape a lesser extent, eastern chestnut mouse site occupancy heterogeneity was affected by patch area (Fig. 5c). Hence, we
accepted H4, landscape structure affected eastern chestnut mouse site occupancy (Table S2; Fig. 4). Species associations ‘Best’ model for predicting eastern chestnut mouse We found a negative effect of bush rat abundance on site occupancy eastern chestnut mouse site occupancy (Fig. 4). There was no significant influence of swamp rat or brown When significant explanatory variables determined antechinus abundance on eastern chestnut mouse site from the univariate HGLMs and all subsets regression occupancy. When we included the occurrence of the were included in a single model, burned 2003 and bush rat, swamp rat and brown antechinus as fixed percent rock cover were the only variables whose terms in a model together, no species had a significant coefficients were not significantly different from zero influence on eastern chestnut mouse site occupancy (Table 5).
123 40 Pereoglou, F., Population processes in an early successional heathland 163 Author's personal copy Appendix 2: Manuscript – Site occupancy Landscape Ecol (2016) 31:255–269 265
Discussion landscape heterogeneity increased. Movement through peripheral habitat can result in direct mortality Importance of landscape structure versus site-level through predation, or indirectly influence survival habitat selection through increased competition for sub-optimal resources (Higgs and Fox 1993; Chesson 2000). As Our study tested the importance of key themes of other vegetation communities replace heathland, or ecological theory from disturbance ecology (site-level late seral stage heath replaces early successional disturbance history), landscape ecology (site-level communities, species that track suitable environmen- habitat quality and landscape patterns) and community tal conditions may become locally extinct (Templeton ecology (interspecific interactions) on the distribution et al. 2011). For example, to avoid eastern chestnut of an early successional species in a fire-prone mouse extinction, early successional heathland would environment. A key finding from our study was that need to become available close to an extant local site occupancy was strongly influenced by landscape population or alternatively, fire itself would need to (i.e. broad-scale landscape structure, hypothesis 4), improve conditions suitable for dispersal (e.g. Tem- and to a lesser extent, elements within the landscape pleton et al. 2011; Banks et al. 2013). Analysis of (i.e. resource availability at a site-level, hypothesis 2) eastern chestnut mouse genetic data supports the latter (Fig. 4). Large heathland habitat patches, a less proposal that recently burned vegetation has greater heterogeneous intervening habitat matrix, and the conductance for gene flow than unburnt habitat presence of dead shrubs indicated a greater likelihood (Pereoglou et al. 2013). The interaction between of sites being occupied by the eastern chestnut mouse. dispersal processes and landscape pattern influences This reflects the species’ known shelter resource the temporal dynamics of populations (Turner 1989). requirements and presumed habitat specialisation. The capacity of individuals of a population to leave The basic principle of landscape ecology is that the one patch to colonize another similar patch is an spatial organization of the environment controls essential process of landscape dynamics after distur- ecological processes (Burel and Baudry 2003). bance (Burel and Baudry 2003). Conservation must Species respond to patch-level variables because they emphasise management of existing habitats and cre- are important predictors of distribution and abundance ation of new habitats within colonising distance (Prugh et al. 2008; Thornton et al. 2011). Habitat (Thomas 1994). specialists that inhabit patchy terrestrial landscapes Species distribution and abundance patterns are could be expected to occur predominantly in large influenced by fine-scale features of the environment habitat patches of the preferred vegetation type and with which they interact most directly (Thornton et al. this was the case in our study. Large habitat patches 2011). Within-patch habitat attributes such as dead can support higher rates of occupancy compared to shrubs were important predictors of site occupancy smaller ones for species in disturbed landscapes and were positively associated with the most recent (Thomas et al. 1992; Banks et al. 2005a). For early major wildfire (Fig. 4). This demonstrates that site successional species, large habitat patches may facil- occupancy of early successional specialists can be itate post-fire population recovery as a result of both facilitated indirectly through the role of fire in creating population expansion from within the fire boundary or maintaining critical habitat features and is indica- (Banks et al. 2011) and colonisation of individuals tive of the eastern chestnut mouse preference for from outside the fire boundary (Watson et al. 2012). sheltering in the collapsed dead shrub layer in recently Properties of the matrix surrounding habitat patches burned heath (Pereoglou et al. 2011). Habitat features are also important (Prugh et al. 2008). The surround- expected to provide shelter have been correlated with ing landscape influences patch dynamics and the animal density for other Pseudomys species—e.g. networks of exchange that exist between habitat Xanthorrhea skirt for the heath rat P. shortridgei (Di patches, facilitating the processes of recolonization Stefano et al. 2011), presence of Banksia ornata (Pickett and White 1985; Burel and Baudry 2003). shrubs for the silky mouse P. apodemoides (Cockburn Therefore, species in isolated patches should have a 1981), and rock cover for the Hastings river mouse P. lower probability of persistence (Turner 1989). In our oralis (Townley 2000). Rock cover was also identified study, patches were less likely to be occupied as in our modelling as an influence on eastern chestnut
123
Pereoglou, F., Population processes in an early successional heathland 164 Author's personal copy Appendix 2: Manuscript – Site occupancy
266 Landscape Ecol (2016) 31:255–269 mouse occupancy. However, unlike the Hastings river and it is possible that fire history and vegetation type mouse, rock crevasses have not previously been were confounding one another in our analyses. Research identified as a shelter resource or a preferred micro- that replicates equal numbers of burned and unburned habitat, so the reasons for this result remain unclear. heathland sites may show a different result. We have The variable rock cover may be confounded with fire shown fire to indirectly affect eastern chestnut mouse history because sites that were not burnt in the 2003 site occupancy by creating critical habitat features used fire had no rock cover. for diurnal refugia. Therefore, the effects of fire may The level of floristic and structural complexity in have been captured entirely by the variable dead shrub. regenerating communities has been linked to small Finally, we may simply not have measured enough fire mammal species assemblages, and underpins assump- attributes to model in our analyses. For example, fire tions on species occurrence in successional ecosys- regimes are possibly more important than fire history in tems (Catling 1991; Masters 1993; Monamy and Fox predicting species occurrence and we were unable to 2010). In our study, floristic richness as well as the distinguish the relative effects of fire frequency, season amount and type of cover were not significant andseverity(Whelan1995; Keith et al. 2002). predictors of eastern chestnut mouse site occupancy (Fig. 4). Our result is consistent with other investiga- Species interactions and the habitat tions, which have uncovered strong species- and site- accommodation model specific variation in habitat preferences (see also Cockburn 1978; Di Stefano et al. 2011). The habitat accommodation model (Fox 1982) has been shown to be a poor predictor of ecological Fire history as a predictor of site occupancy patterns in relation to fire in other systems (Letnic et al. for an early successional specialist 2004; Driscoll and Henderson 2008; Di Stefano et al. 2011). The absence of predictable sympatric small Natural disturbances leave legacies that persist for mammal species effects on eastern chestnut mouse site decades to centuries and integrating this history with occupancy (Fig. 4) suggested competitive exclusion current understanding of landscape patterns and was not the strongest biological mechanism structur- processes remains an important goal of landscape ing the small mammal community as predicted in a ecology (Turner 2005). A key finding of this study was mammalian secondary succession (Fox 1982). There that eastern chestnut mouse site occupancy was not was some indication that site occupancy of the eastern directly affected by fire history (hypothesis 1) (Fig. 4). chestnut mouse increased in the absence of the bush rat Potential explanations for an absence of direct fire but this effect disappeared when all co-occurring history effects are two-fold. First, by comparison with species were modelled together. In other studies, the the common co-occurring small mammal species departure from theoretical predictions has been which showed negative responses to fire, the ability attributed to various processes including the absence to endure disturbance seems to be a necessary trait of competitors (Catling 1986; Di Stefano et al. 2011), amongst habitat specialists (Hutto 2008). In another rapid recovery of vegetation (Lindenmayer et al. study, where researchers experimentally simulated the 2008b, 2009) and correlation with alternative explana- effects of fire, populations of the eastern chestnut tory variables (Letnic et al. 2004; Driscoll and mouse were also unaffected (Fox et al. 2003). Henderson 2008). However, in situ survival of indi- Individuals of the eastern chestnut mouse that survive viduals during a fire event probably has a greater after fire could be considered ‘biological legacies’ influence on population persistence after fire (Linden- (sensu Franklin et al. 2000). In another investigation in mayer et al. 2005; Banks et al. 2011) compared to our study region, eastern bristlebirds (Dasyornis relative thresholds of habitat suitability. Therefore, it brachypterus) that persisted in burned sites also were seems unrealistic to expect theoretical models to considered to be survivors of a pre-fire fauna (Lin- accurately forecast species assemblages, as they are denmayer et al. 2009). likely to vary with the complexities of disturbance Secondly, the effects of fire history were either regimes, spatial patterns of landscape continuity, site- captured by, or confounded with other variables. Eighty level habitat features, and the interactions among these percent of heathland sites were burned in the 2003 fire, factors. 45 123
Pereoglou, F., Population processes in an early successional heathland 165 Author's personal copy Appendix 2: Manuscript – Site occupancy Landscape Ecol (2016) 31:255–269 267
Conclusions Banks SC, Ward SJ, Lindenmayer DB, Finlayson GR, Lawson SJ, Taylor AC (2005b) The effects of habitat fragmentation on the social kin structure and mating system of the agile Our study evaluates the effects of disturbance, spatial antechinus, Antechinus agilis. Mol Ecol 14:1789–1801 patterns of landscape structure and site-level habitat Banks SC, Dujardin M, McBurney L, Blair D, Barker M, Lin- features on species occupancy in an early successional denmayer DB (2011) Starting points for small mammal community. We found a positive effect of heathland population recovery after wildfire: recolonisation or residual populations? Oikos 120:26–37 patch area and a negative effect of landscape hetero- Banks SC, Cary GJ, Smith AL, Driscoll DA, Lindenmayer DB, geneity on site occupancy by the eastern chestnut Davies ID, Gill AM, Peakall R (2013) How does ecological mouse. Eastern chestnut mouse site occupancy also disturbance influence genetic diversity? Trends Ecol Evol was associated with habitat features, such as a 28:670–679 Bradstock RA, Williams J, Gill AM (eds) (2012) Flammable collapsed dead shrub layer, that may be created by Australia: the fire regimes and biodiversity of a continent. burning. We found recent fire and the effect of Cambridge University Press, Cambridge multiple fires per se did not have a strong influence on Braithwaite RW (1982) Small marginal groups of Rattus eastern chestnut mouse site occupancy. We recom- lutreolus in good and poor habitat. J Zool 198:529–532 Breed B, Ford F (2007) Native mice and rats. CSIRO Publish- mend surveys targeting the species take account of ing, Collingwood landscape configuration for optimal results. Further- Burel F, Baudry J (2003) Landscape ecology: concepts, methods more, fire regimes expected to promote eastern and applications. Science Publishers, Enfield chestnut mouse population growth should encourage Burnham KP, Anderson DR (2002) Model selection and multi- modal inference: a practical information-theoretic the retention of critical habitat features rather than be approach. Springer, New York based entirely on temporal rates of successional Catling PC (1986) Rattus lutreolus, colonizer of heathland after stages. We confirmed the general importance of the fire in the absence of Pseudomys species. Aust Wildl Res multi-level approach to landscape ecology studies, as 13:127–139 Catling PC (1991) Ecological effects of prescribed burning our study species responded to landscape, patch, and practices on the mammals of southeastern Australia. In: within-patch variables (Thornton et al. 2011). Our Lunney D (ed) Conservation of Australia’s forest fauna. work suggests that a multi-scaled approach to inves- Royal Zoological Society of NSW, NSW, pp 353–363 tigations of site occupancy should be taken to ensure Chesson P (2000) Mechanisms of maintenance of species diversity. Annu Rev Ecol Syst 31:343–366 species- and site-specific resource requirements are Cockburn A (1978) Distribution of Pseudomys shortridgei identified and that the impacts of changes to distur- (Muridae: Rodentia) and its relevance to that of other bance regimes on these resources can be effectively heathland Pseudomys. Aust Wildl Res 5:213–219 evaluated. Cockburn A (1981) Diet and habitat preference of the silky desert mouse Pseudomys apodemoides (Rodentia). Aust Wildl Res 8:475–497 Acknowledgments We most gratefully acknowledge the Cockburn A, Lazenby-Cohen KA (1992) Use of nest trees by Wreck Bay Aboriginal Community and the Wandandian Antechinus stuartii, a semelparous lekking marsupial. People of the Dharawal and Dhurga language groups on J Zool London 226:657–680 whose lands this study was conducted. Mason Crane, Damien Cockburn A, Braithwaite RW, Lee AK (1981) The response of Michael, Rebecca Montague-Drake, and Sachiko Okada the heath rat, Pseudomys shortridgei, to pyric succession— provided fieldwork assistance. This project is part of a major a temporally dynamic life-history strategy. J Anim Ecol study in Booderee National Park funded by the Australian 50:649–666 Research Council, the Department of Environment and Water Connell JH, Slatyer RO (1977) Mechanisms of succession in Resources and the Department of Defence. natural communities and their role in community stability and organization. Am Nat 111:1119–1144 Conover WJ (1999) Practical nonparametric statistics, 3rd edn. Wiley, New York References Crowther MS (2002) Distributions of species of the Antechinus stuartii-A. flavipes complex as predicted by bioclimatic Baker J, Whelan RJ, Evans L, Moore S, Norton M (2010) modelling. Aust J Zool 50:77–91 Managing the ground parrot in its fiery habitat in south- Crowther MS, Braithwaite RW (2008) Brown antechinus, An- eastern Australia. Emu 110:279–284 techinus stuartii. In: Van Dyck S, Strahan R (eds) The Banks SC, Finlayson GR, Lawson SJ, Lindenmayer DB, Paet- mammals of Australia. Reed New Holland, Sydney, kau D, Ward SJ, Taylor AC (2005a) The effects of habitat pp 94–96 fragmentation due to forestry plantation establishment on Delmoral R, Bliss LC (1993) Mechanisms of primary succes- the demography and genetic variation of a marsupial car- sion: insights resulting from the eruption of Mount St nivore, Antechinus agilis. Biol Conserv 122:581–597 Helens. Adv Ecol Res 24:1–66
123
Pereoglou, F., Population processes in an early successional heathland 166 Author's personal copy Appendix 2: Manuscript – Site occupancy
268 Landscape Ecol (2016) 31:255–269
Di Stefano J, Owen L, Morris R, Duff T, York A (2011) Fire, fragments following major population reduction. J Appl landscape change and models of small mammal habitat Ecol 42:649–658 suitability at multiple spatial scales. Aust Ecol 36:638–649 Lindenmayer DB, MacGregor C, Welsh A, Donnelly C, Crane Dickman CR (2008) Agile antechinus, Antechinus agilis. In: M, Michael D, Montague-Drake R, Cunningham RB, Van Dyck S, Strahan R (eds) The mammals of Australia. Brown D, Fortescue M, Dexter N, Hudson M, Gill AM Reed New Holland, Sydney, pp 83–84 (2008a) Contrasting mammal responses to vegetation type Driscoll DA, Henderson MK (2008) How many common reptile and fire. Wildl Res 35:395–408 species are fire specialists? A replicated natural experiment Lindenmayer DB, Wood JT, MacGregor C, Michael DR, Cun- highlights the predictive weakness of a fire succession ningham RB, Crane M, Montague-Drake R, Brown D, model. Biol Conserv 141:460–471 Muntz R, Driscoll DA (2008b) How predictable are reptile ESRI 2009. ArcMap 9.2. Environmental Systems Resource responses to wildfire? Oikos 117:1086–1097 Institute, California Lindenmayer DB, MacGregor C, Wood JT, Cunningham RB, Fox BJ (1982) Fire and mammalian secondary succession in an Crane M, Michael D, Montague-Drake R, Brown D, Australian coastal heath. Ecology 63:1332–1341 Fortescue M, Dexter N, Hudson M, Gill AM (2009) What Fox BJ (2008) Eastern chestnut mouse, Pseudomys gracilicau- factors influence rapid post-fire site re-occupancy? A case datus. In: Van Dyck S, Strahan R (eds) The mammals of study of the endangered eastern bristlebird in eastern Australia. Reed New Holland, Sydney, pp 634–635 Australia. Intern J Wildl Fire 18:84–95 Fox BJ, Archer E (1984) The diets of Sminthopsis murina and Lunney D (2008a) Bush rat, Rattus fuscipes. In: Van Dyck S, Antechinus stuartii (Marsupialia: Dasyuridae) in sympatry. Strahan R (eds) The mammals of Australia. Reed New Aust Wildl Res 11:235–248 Holland, Sydney, pp 685–687 Fox BJ, Kemper CM (1982) Growth and development of Lunney D (2008b) Swamp rat, Rattus lutreolus. In: Van Dyck S, Pseudomys gracilicaudatus (Rodentia: Muridae) in the Strahan R (eds) The mammals of Australia. Reed New laboratory. Aust J Zool 30:175–185 Holland, Sydney, pp 690–692 Fox BJ, Taylor JE, Thompson PT (2003) Experimental manip- Luo JIA, Fox BJ (1996) Seasonal and successional dietary shifts ulation of habitat structure: a retrogression of the small of two sympatric rodents in coastal heathland: a possible mammal succession. J Anim Ecol 72:927–940 mechanism for coexistence. Aust Ecol 21:121–132 Franklin JF, Lindenmayer D, MacMahon JA, McKee A, Mag- Luo J, Fox BJ, Jefferys E (1994) Diet of the Eastern Chestnut nuson J, Perry DA, Waide R, Foster D (2000) Threads of Mouse (Pseudomys gracilicaudatus) I composition, continuity. Conserv Pract 1:8–17 diversity and individual variation. Wildl Res 21:401–417 Gill AM, Allan G (2008) Large fires, fire effects and the fire- MacArthur RH, Wilson EO (1967) The theory of island bio- regime concept. Intern J Wildl Fire 17:688–695 geography. Princeton University Press, New Jersey Higgs P, Fox BJ (1993) Interspecific competition: a mechanism Masters P (1993) The effects of fire driven succession and for rodent succession after fire in wet heathland. Aust J rainfall on small mammals in spinifex grassland at Ecol 18:193–201 Uluru National Park, Northern Territory. Wildl Res Hutto RL (2008) The ecological importance of severe wildfires: 20:803–813 some like it hot. Ecol Appl 18:1827–1834 McCarthy MA, Gill AM, Lindenmayer DB (1999) Fire regimes Keith DA, McCaw WL, Whelan RJ (2002) Fire regimes in in mountain ash forest: evidence from forest age structure, Australian heathlands and their effects on plants and ani- extinction models and wildlife habitat. For Ecol Manage mals. In: Bradstock RA, Williams JE, Gill AM (eds) 124:193–203 Flammable Australia: the fire regimes and biodiversity of a McCullagh P, Nelder JA (1989) Generalized linear models. continent. Cambridge University Press, New York, Chapman and Hall, New York pp 199–237 Monamy V, Fox BJ (2000) Small mammal succession is Lee Y, Nelder JA, Pawitan Y (2006) Generalized linear models determined by vegetation density rather than time elapsed with random effects: unified analysis via h-likelihood. since disturbance. Aust Ecol 25:580–587 Chapman & Hall/CRC, Boca Raton Monamy V, Fox BJ (2010) Responses of two species of Letnic M, Dickman CR (2005) The responses of small mammals heathland rodents to habitat manipulation: vegetation to patches regenerating after fire and rainfall in the Simp- density thresholds and the habitat accommodation model. son Desert, central Australia. Aust Ecol 30:24–39 Aust Ecol 35:334–347 Letnic M, Dickman CR, Tischler MK, Tamayo B, Beh CL Newsome AE, McIlroy J, Catling PC (1975) The effects of an (2004) The responses of small mammals and lizards to extensive wildfire on populations of twenty ground verte- post-fire succession and rainfall in arid Australia. J Arid brates in southeast Australia. Proc Ecol Soc Aust Environ 59:85–114 9:107–123 Letnic M, Tischler M, Gordon C (2013) Desert small mammal Pereoglou F, MacGregor C, Banks SC, Ford FD, Wood JT, responses to wildfire and predation in the aftermath of a La Lindenmayer DB (2011) Refuge site selection by the Nina driven resource pulse. Aust Ecol 38:841 eastern chestnut mouse in recently burned heath. Wildl Res Levin SA (1992) The problem of pattern and scale in ecology: 38:290–298 the Robert H. MacArthur award lecture. Ecology Pereoglou F, Lindenmayer DB, MacGregor C, Ford F, Wood J, 73:1943–1967 Banks SC (2013) Landscape genetics of an early succes- Lindenmayer DB, Cunningham RB, Peakall R (2005) The sional specialist in a disturbance-prone environment. Mol recovery of populations of bush rat Rattus fuscipes in forest Ecol 22:1267–1281 50
123
Pereoglou, F., Population processes in an early successional heathland 167 Appendix 2: Manuscript – Site occupancy
Author's personal copy
Landscape Ecol (2016) 31:255–269 269
Peterson GD (2002) Contagious disturbance, ecological mem- Thomas CD, Thomas JA, Warren MS (1992) Distributions of ory, and the emergence of landscape pattern. Ecosystems occupied and vacant butterfly habitats in fragmented 5:329–338 landscapes. Oecologia 92:563–567 Pickett STA, White PS (eds) (1985) The ecology of natural Thornton DH, Branch LC, Sunquist ME (2011) The influence of disturbance and patch dynamics. Academic Press Inc, landscape, patch, and with-in patch factors on species Florida presence and abundance: a review of focal patch studies. Prugh LR, Hodges KE, Sinclair ARE, Brashares JS (2008) Landscape Ecol 26:7–18 Effect of habitat area and isolation on fragmented animal Townley SJ (2000) The ecology of the Hastings River mouse populations. Proc Natl Acad Sci 105:20770–20775 Pseudomys oralis (Rodentia: Muridae) in northeastern Robinson AC (1987) The ecology of the bush rat, Rattus fus- New South Wales and southeastern Queensland. PhD, cipes (Rodentia: Muridae), in Sherbrooke Forest, Victoria. Southern Cross University, Lismore Aust Mammal 11:35–49 Turner MG (1989) Landscape ecology: the effect of pattern on Robinson NM, Leonard SWJ, Ritchie EG, Bassett M, Chia EK, process. Annu Rev Ecol Syst 20:171–197 Buckingham S, Gibb H, Bennett AF, Clarke MF (2013) Turner MG (2005) Landscape ecology: what is the state of the Refuges for fauna in fire-prone landscapes: their ecological science. Annu Rev Ecol Syst 36:319–344 function and importance. J Appl Ecol 50:1321–1329 Turner MG, Baker WL, Peterson CJ, Peet RK (1998) Factors Rossi F, Forster RM, Montserrat F, Ponti M, Terlizzi A, Yse- influencing succession: lessons from large, infrequent baert T, Middelburg JJ (2007) Human trampling as short- natural disturbances. Ecosystems 1:511–523 term disturbance on intertidal mudflats: effects on macro- Velle LG, Vandvik V (2014) Succession after prescribed fauna biodiversity and population dynamics of bivalves. burning in coastal Calluna heathlands along a 340-km Mar Biol 151:2077–2090 latitudinal gradient. J Veg Sci 25:546–558 Rota CT, Millspaugh JJ, Rumble MA, Lehman CP, Kesler DC Vieira EM, Briani DC (2013) Short-term effects of fire on small (2014) The role of wildfire, prescribed fire, and mountain rodents in the Brazillian Cerrad and their relation with pine beetle infestations on the population dynamics of feeding habits. International Journal of Wildland Fire. black-backed woodpeckers in the Black Hills, South 22:1063–1071 Dakota. PLoS ONE 9:e94700 Warneke RM (1971) Field study of the bush rat (Rattus fus- Schwarz G (1978) Estimating the dimension of a model. Ann cipes). Wildl Contrib Vic 14:1–115 Stat 6:461–464 Watson SJ, Taylor RS, Nimmo DG, Kelly LT, Clarke MF, Sousa WP (1979) Disturbance in intertidal boulder fields: the Bennett AF (2012) The influence of unburnt patches and non-equilibrium maintenance of species diversity. Ecology distance from refuges on post-fire bird communities. Anim 60:1225–1239 Conserv 15:499–507 Specht RL (1981) Responses to fire in heathlands and related Watts CHS, Aslin HJ (1981) The rodents of Australia. Angus & shrublands. In: Gill AM, Groves RH, Noble IR (eds) Fire Robertson Publishers, Sydney and the Australian biota. Australian Academy of Science, Whelan RJ (1995) The ecology of fire. Cambridge University Canberra, pp 395–415 Press, Cambridge Swanson ME, Franklin JF, Beschta RL, Crisafulli CM, Del- Whelan RJ, Rodgerson L, Dickman CR, Sutherland EF (2002) laSala DA, Hutto RL, Lindenmayer DB, Swanson FJ Critical life cycles of plants and animals: developing a (2011) The forgotten stage of forest succession: early- process-based understanding of population changes in fire- successional ecosystems on forest sites. Front Ecol Environ prone landscapes. In: Bradstock RA, Williams JE, Gill AM 9:117–125 (eds) Flammable Australia: the fire regimes and biodiver- Taws N (1997) Vegetation survey and mapping of Jervis Bay sity of a continent. Cambridge University Press, New York, Territory. Report, Canberra pp 94–124 Templeton AR, Brazeal H, Neuwald JL (2011) The transition Williams DG (1995) Heaths and scrubs. In: Cho G, Georges A, from isolated patches to a metapopulation in the eastern Stoutjesdijk R, Longmore R (eds) Jervis Bay. A place of collared lizard in response to prescribed fires. Ecology cultural, scientific and educational value. Kowari 5. Aus- 92:1736–1747 tralian Nature Conservation Agency, Canberra, pp 83–90 Thomas CD (1994) Extinction, colonization, and metapopula- tions: environmental tracking by rare species. Conserv Biol 8:373–378
55
Pereoglou, F., Population processes in an early successional heathland 168 123
Appendix 3: Manuscript – Landscape genetics
Manuscript: Landscape genetics in a disturbance-prone environment Published, March 2013
Pereoglou, F., Population processes in an early successional heathland 169 Appendix 3: Manuscript – Landscape genetics
Molecular Ecology (2013) 22, 1267–1281 doi: 10.1111/mec.12172
Landscape genetics of an early successional specialist in a disturbance-prone environment
F. PEREOGLOU,* D. B. LINDENMAYER,* C. MACGREGOR,* F. FORD,† J. WOOD* and S. C. BANKS* *Fenner School of Environment and Society, The Australian National University, Canberra, ACT 0200, Australia, †Directorate of Heritage and Biodiversity Conservation, Department of Defence Support Group, Brindabella Business Park, PO Box 7925, Canberra BC, ACT 2610, Australia
Abstract Species that specialize in disturbed habitats may have considerably different dispersal strategies than those adapted to more stable environments. However, little is known of the dispersal patterns and population structure of such species. This information is important for conservation because many postfire specialists are at risk from anthropo- genic changes to natural disturbance regimes. We used microsatellite markers to assess the effect of landscape variation and recent disturbance history on dispersal by a small mammal species that occupies the early seral stage of vegetation regeneration in burnt environments. We predicted that a postfire specialist would be able to disperse over multiple habitat types (generalist) and not exhibit sex-biased dispersal, as such strate- gies should enable effective colonization of spatially and temporally variable habitat. We found significant differentiation between sites that fitted an isolation-by-distance pattern and spatial autocorrelation of multilocus genotypes to a distance of 2–3 km. There was no consistent genetic evidence for sex-biased dispersal. We tested the influ- ence of different habitat- and fire-specific landscape resistance scenarios on genetic dis- tance between individuals and found a significant effect of fire. Our genetic data supported recently burned vegetation having greater conductance for gene flow than unburnt habitat, but variation in habitat quality between vegetation types and occu- pied patches had no effect on gene flow. Postfire specialists must evolve an effective dispersal ability to move over distances that would ensure access to early successional stage vegetation. Natural disturbance and natural heterogeneity may therefore not influence population genetic structure as negatively as expected.
Keywords: dispersal, fire, habitat heterogeneity, landscape resistance Received 6 July 2011; revision received 20 October 2012; accepted 4 November 2012
abundance (Fahrig & Paloheimo 1988) and habitat selec- Introduction tion (Morris 1992). Recently, landscape genetics research Dispersal behaviour and landscape configuration are has contributed to our understanding of population primary determinants of population genetic structure connectivity in relation to landscape elements (Manel (Bowler & Benton 2005; Baguette & Van Dyck 2007). et al. 2003; Storfer et al. 2010). However, threatened The landscape matrix can facilitate or impede move- species with habitat preferences for disturbed envi- ment of organisms between populations (Taylor et al. ronments are not well represented in the population 1993; Lindenmayer & Fischer 2006; Fahrig 2007) with genetics literature. The ecological requirements of these dispersal behaviour influencing spatial patterns in specialist species suggest that their patterns of dispersal and demography differ considerably from species that utilize more stable environments and that this is
Correspondence: Felicia Pereoglou, Fax: +61 2 6125 0746; likely to be revealed in patterns of genetic diversity and E-mail: [email protected] structure.
© 2013 Blackwell Publishing Ltd Pereoglou, F., Population processes in an early successional heathland 170 Appendix 3: Manuscript – Landscape genetics
1268 F. PEREOGLOU ET AL.
In conjunction with landscape data, spatial genetic existing populations and colonization of new areas information can identify habitat variables that maintain (particularly with a large number of colonists) are connectivity among populations or serve as barriers to expected to weaken genetic structure (Wade & animal movements (Scribner et al. 2005). Current McCauley 1988; Templeton et al. 2001; Excoffier et al. research on landscape connectivity has focussed on the 2009). Our prediction may be incorrect if the number influence of ‘habitat breaks’ (i.e. anthropogenic land- of individuals colonizing newly suitable habitat after scape fragmentation; Gauffre et al. 2008; Macqueen et al. fire is very small and ongoing migration in subse- 2008; Spear & Storfer 2008; Pavlacky et al. 2009), natural quent generations is rare (Wade & McCauley 1988). barriers (e.g. rivers, mountains and topography; Spear However, our hypothesis was that the former predic- et al. 2005; Murphy et al. 2010) or a combination of both tion would hold. (Cushman et al. 2006; Perez-Espona et al. 2008; Short 2 Natural habitat heterogeneity will not restrict gene Bull et al. 2011). There is limited information on the flow. Habitat suitability has been considered as a effect of natural disturbance such as wildfire, naturally correlate of ‘conductance’ to dispersal across a land- heterogeneous landscapes or natural variation in habitat scape, on the assumption that environmental types suitability on patterns of genetic diversity, population that rank poorly by occupancy or abundance criteria structure and gene flow. will also be poor for dispersal (Wang et al. 2008; Fire is a major form of natural disturbance (Turner Spear et al. 2010). The eastern chestnut mouse has a et al. 1998). Different fire regimes produce distinct land- naturally patchy distribution within our study area scape patterns and species arrays (Bond & Keeley 2005). due to very specific requirements for vegetation type Population recovery and persistence in postfire environ- and fire history (Pereoglou et al. 2011). Habitat suit- ments are influenced by fire regimes and their effects ability is thus spatially patchy and temporally on landscape features including patch isolation and dynamic. We predicted that an effective dispersal connectivity (Clarke 2008). With altered fire regimes strategy to access and colonize suitable habitat under and increased fire occurrence being a likely impact of such dynamic environmental conditions would be climate change (Williams et al. 2001; Flannigan et al. one in which dispersal is possible across all landscape 2009; Steffen et al. 2009), our understanding of how fire types and is not restricted by perceived habitat events influence biota is a critical contemporary preferences. environmental issue. Fire playing a positive role in 3 Dispersal will not be strongly sex biased. Male-biased maintaining habitat suitability is an effect that is plausi- dispersal is common among mammals (Greenwood ble for postfire specialist fauna and those that colonize 1980) and occurs in other small mammal species that early successional ecosystems, but one that is not well occur in our study area, including the brown antechi- studied. nus (Antechinus stuartii) and the bush rat (Rattus fusci- In this study, we examined the population genetic pes; Cockburn et al. 1985; Peakall et al. 2003). structure of the eastern chestnut mouse (Pseudomys However, we predicted a lack of strong sex bias in gracilicaudatus Gould 1845), a threatened small mammal dispersal for an early seral specialist. Modelling sug- endemic to the east coast of Australia and dependent gests that strong sex-biased dispersal is not favoured on regular fire. We addressed a series of hypotheses under high demographic stochasticity (Gros et al. relating to the dispersal strategy of a postdisturbance 2008). After fire, dispersal by eastern chestnut mice is specialist mammal, including the following: likely to be associated with colonization of newly suitable habitat, and greater colonization ability by 1 Fire/burned areas will promote gene flow. Several males compared with females would be disadvanta- studies have shown that fire affects genetic structure geous in that it would increase the likelihood of set- through environmental changes that cause demo- tling in unoccupied habitat with no potential mates. graphic variation such as population expansion (including colonization) or decline and variation in connectivity (Templeton et al. 2001; Schrey et al. 2010, Methods 2011; Templeton et al. 2011). For species that special- ize in early seral stage, postdisturbance habitats, we Study species expect a reduction in genetic structure in recently burned areas. This is because fire increases the area Pseudomyine rodents provide some of the best exam- of suitable habitat, so small and relatively isolated ples of fire specialist fauna (Catling & Newsome 1981) prefire populations will expand and individuals due to their rapid establishment and early breeding will disperse to colonize newly suitable habitat. Pop- after fire, rapid population increase and ability to main- ulation expansion, increased connectivity between tain viable breeding populations from first colonization
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 171 Appendix 3: Manuscript – Landscape genetics
LANDSCAPE GENETICS OF A POSTFIRE SPECIALIST 1269
(Fox 1982). The eastern chestnut mouse is an early seral (Fig. 1). We genotyped 287 individual animals, and specialist that colonizes heathland habitat, peaking in sample sizes ranged from 1 to 58 mice per site (Fig. 1). abundance postfire before declining over time as vege- We used two approaches to survey the study area. Ini- tation regenerates (Fox 1982; Monamy & Fox 2000; Fox tial site selection was randomly stratified by vegetation et al. 2003). It has a disjunct distribution along the east type and consisted of a single 100-m transect. Each tran- coast of Australia and occurs at low density throughout sect comprised the following: (i) a black plastic drift much of its range. The species was considered extinct fence connecting 380 mm deep 9 300 mm diameter in southern Australia until 2002 when it was discovered pitfall traps at 0–20, 40–60 and 80–100 m; (ii) large wire in the Jervis Bay region (Meek & Triggs 1997). This area cage traps (30 9 30 9 60 cm) placed at 0 and 100 m now supports the only contemporary population of the with four small wire cage traps (20 9 20 9 50 cm) at eastern chestnut mouse south of Gosford, New South 20 m intervals between; and (iii) ten aluminium box Wales (33°23′S, 151°18′E) and is arguably the most traps (Elliot Scientific Equipment, Australia) placed isolated population of the species in eastern Australia. every 10 m along the transect line from 0 to 90 m. Data The eastern chestnut mouse is an annual summer on the number of transects in each vegetation type, the breeder capable of producing multiple litters of one to number with eastern chestnut mouse captured and the five (usually three) offspring (Fox 2008). The species corresponding occupancy estimates are given in lives for <2 years in wild populations (Watts & Kemper Table S1 (Supporting information). We then chose five 1989). Early development is rapid, which contrasts with transects and selected an additional four sites for inten- a slow growth rate to reach maturity (Fox & Kemper sive grid trapping to sample sufficiently high densities 1982). Adults show significant sexual dimorphism (Fox of the eastern chestnut mouse to allow for population & Kemper 1982) and attain average weights of 90 g for genetic analyses. We classified all grid sites as dry males and 75 g for females (Fox 2008). Little is known heath (Taws 1997), and each had been burned by wild- about social organization and behaviour in this species. fire in December 2003. Our grid arrangement comprised Radio-tracking data suggest that adult eastern chestnut lines at 20 m intervals with traps 10 m apart covering mice are solitary, territorial and maintain female home an area ranging from 1 to 3 ha (1.9 0.19 ha). Transect ranges of 0.5–1 ha and up to 4 ha for males (F. Pereog- sites were surveyed annually during the eastern chest- lou unpublished data). nut mouse breeding season (November–February), and grid sites were surveyed twice each year, during the breeding season and also during the nonbreeding sea- Study area and sampling son (June–August). We opened traps for three or four We collected tissue from 26 sites surveyed between consecutive nights (31 548 trap nights in total) and bai- 2008 and 2010 in Booderee National Park, located ted with rolled oats, peanut butter and fennel seed. We approximately 200 km south of Sydney, Australia assessed captured animals for weight, sex, age cohort
Fig. 1 Study area and distribution of sur- vey sites. Solid symbol indicates site with eastern chestnut mouse presence, and the number of genotyped individuals is pro- vided above symbol. Hollow symbol indicate sites with no eastern chestnut mouse captures.
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 172 Appendix 3: Manuscript – Landscape genetics
1270 F. PEREOGLOU ET AL.
and breeding condition and collected a 2-mm ear calculated pairwise values of FST for all sites where biopsy that is stored in 100% ethanol. sample sizes were >10 individuals. We used Mantel
tests to test for isolation by distance on pairwise FST between sites (F /1 F vs. log geographic distance). DNA extraction, PCR amplification and genotyping ST ST Unless otherwise stated, all permutational tests were We extracted DNA from ear tissue using the method of performed 999 times, and analyses were completed (Bruford et al. 1988) and re-suspended the DNA pellet using GENALEX. in 100 lL of TE buffer (10 mM Tris and 0.1 mM EDTA). To further investigate the spatial pattern of genetic We genotyped samples at 10 microsatellite loci (Moro & structure, we calculated the spatial autocorrelation (r)of Spencer 2003; Table S2, Supporting information). Micro- pairwise multilocus genetic distances (GD) over a set of satellite PCRs with a total volume of 30 lL contained geographic distance classes (Smouse & Peakall 1999). 9 l 10 buffer (Qiagen), 200 M dNTPs, 1.8 mM MgCl2, We calculated pairwise multilocus GD according to the 0.5 lL BSA, 4 pmoles reverse primer, 1.5 pmoles for- methods of Smouse & Peakall (1999) and tested the null ward-tailed primer, 2.5 pmoles fluorescent end-labelled hypothesis of no spatial genetic pattern (r = 0) against M13 primer sequence (Schuelke 2000), 0.2 units Taq the alternative that spatial autocorrelation was positive polymerase (Qiagen) and 1.5 lL template DNA. We ran at some distances and negative at others (r 6¼ 0). For all reactions on a thermal cycler (Applied Biosystems, each survey year (2008–2009 n = 172, 2009–2010 Inc.) with initial denaturation of 94–95 °C for 3 min, n = 141), we ran two different analyses. In the first then cycled using PCR conditions described in Table S2 analysis, we selected 500-m distance classes up to 5 km. (Supporting information), followed by a final extension The second was a finer-scaled analysis, where we inves- step of 72 °C for 10 min. We ran the PCR products on tigated spatial autocorrelation at 50-m intervals up to an ABI 3130 (Applied Biosystems Inc.) automated 250 m (i.e. fine-scale structure within sampling sites). sequencer with the LIZ 500 size standard. We ran nega- The choice of distance classes primarily relates to our tive controls in PCRs and in each sequencer run to sampling scheme, where the first (coarse-scaled) dis- check for contamination. We used GeneMapper version tance classes were larger than our sampling sites, so we 3.7 (Applied Biosystems Inc.) to score each locus for were testing differences between spatial autocorrelation each sample. among individuals within sites (0–500 m) and at increasing distances between sites. The finer-scaled analysis focussed on testing for spatial clustering of Genetic data analysis similar genotypes within the scale of our sampling We used the genotype matching function of MSTOOLS grids. We used 999 permutations with a bootstrap (Park 2001) to identify identical genotypes. Duplicate re-sampling procedure run 1000 times to estimate the genotypes were re-labelled if the corresponding capture 95% confidence interval. records were consistent with multiple captures and We used the programs Structure version 2.3 (Pritchard samples from the same individual (e.g. due to tempo- et al. 2000) and GENELAND version 3.2.4 (Guillot et al. rary marking or tag loss). We ran analyses separately 2005) to complement our other analyses of population for each survey year. Survey years included a 12-month structure with the specific aim of testing whether our period from the beginning of the eastern chestnut samples could be clustered into spatially discrete sub- mouse breeding season in September to August of the populations (parameters in Appendix S1, Supporting following year (September–August 2008–2009 and information). September–August 2009–2010). We used the program GESTE v2 (Foll & Gaggiotti 2006) to estimate effects of landscape variables on population-
Population genetic summary statistics. We excluded specific FST. For this analysis, we restricted data set to locations from the population-level analyses if sites had nine sites with the greatest sample sizes, from which a <10 individuals genotyped. We calculated standard mean of 28 individuals was sampled (SD = 16). All of population genetic summary statistics for all loci and these sites were within areas burnt in 2003, so we were sites including the number of alleles, observed and unable to test for effects of fire history. However, we expected heterozygosities and FIS using GENALEX tested for effects of vegetation heterogeneity with version 6.3 (Peakall & Smouse 2006). We tested for variables representing the proportion of heath (the most departures from Hardy–Weinberg (H–W) equilibrium preferred vegetation type) within 200 and 500 m using FSTAT (Goudet 2001). buffers, as well as sample size (as a proxy for local pop- ulation density) and X and Y coordinates to test for Population subdivision and isolation by distance (IBD). We broad spatial gradients in the strength of genetic – used AMOVA to estimate population subdivision and structure (e.g. an east west peninsula effect). For this
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 173 Appendix 3: Manuscript – Landscape genetics
LANDSCAPE GENETICS OF A POSTFIRE SPECIALIST 1271 analysis, we pooled individuals from both years within with user-defined dimensions. We used the pairwise the same sites. Many individuals were sampled in both source/ground modelling mode to calculate connectiv- years, and previous analyses showed no major differ- ity between genotype point locations with habitat ences in the patterns detected in either year. data specifying per cell conductance values, an eight- neighbour cell connection and calculations based on Sex-biased dispersal. To test the hypothesis that genetic average conductance. structure and corresponding philopatry are not sex We assigned conductance values to raster cells in biased, we used two approaches. First, we used the ARCGIS version 9.2 (ESRI 2009) representing hypotheses heterogeneity test of Smouse et al. (2008) to compare the about the influence of distance, fire and vegetation type spatial autocorrelation r-values of males and females on genetic connectivity (Fig. 2; Table 1). with the multipopulations procedure in GENALEX.We assessed spatial autocorrelation at two scales: 500-m 1 Distance: Our first hypothesis and simplest model distance classes up to 5-km and 50-m distance classes was a test of isolation by distance across a uniform- up to 250 m. If there was a sex bias, for the philopatric resistance terrestrial landscape (Wright 1943). We did sex we expected declining correlations of r-values with not use the classic Euclidean distance measure as it increasing distance if the null hypothesis of r = 0 was was inappropriate to represent straight line distances, not correct. If both sexes showed a similar pattern of given the study area was a peninsula and the coast- decline, we expected the r-values in the first distance line and ocean create a substantial barrier for a terres- would be largest for the philopatric sex (Smouse et al. trial small mammal. Therefore, in our null model, we 2008). We focussed on the first distance class because if assumed movement could occur in any direction, all one sex remains relatively philopatric (such that rela- raster cell values were equal in conductance, and the tives are spatially clustered), any differences in spatial only limit to gene flow was that imposed by the autocorrelation between the sexes are expected to be coastline and ocean. most apparent in this class (Banks & Peakall 2012). We 2 Fire: Our second hypothesis was that fire promotes used 999 permutations with a bootstrap re-sampling genetic connectivity for fauna that specialize in early procedure run 1000 times to estimate the 95% confi- successional ecosystems. We expected to find a posi- dence interval. Sample size differed by survey year: tive effect of fire (i.e. a relative negative effect of 2008–2009 male n = 79, female n = 90 and 2009–2010 unburnt areas) because it resets the vegetation succes- male n = 65, female n = 76. Second, we used a set of sion creating suitable habitat for postfire specialist one-sided biased dispersal tests in FSTAT (Goudet et al. species (Monamy & Fox 2000). Residual survivors 2002). We permuted these tests 10 000 times to compare become responsible for postperturbation population the estimated metrics including mean corrected assign- recovery (Turner et al. 1998; Banks et al. 2011) poten- ment index, variance of corrected assignment index, tially following burned vegetation to recolonize and
FST, FIS and HS between male and female animals. reproduce. Because causal modelling provides only Sample size differed by survey year: 2008–2009 n = 160 significance tests and does not yield estimates of animals in 12 sites and 2009–2010 n = 132 animals in 10 resistance from which we could assess the effect of sites. Sites without adequate sample sizes of both sexes fire, we took the approach of testing a series of mod- were excluded from analysis in FSTAT. els covering what we considered to be the range of plausible conductance values of fire (Table 1). Landscape resistance. We used partial Mantel tests in a 3 Vegetation type: Our third hypothesis was that for causal modelling framework to analyse the influence of species with spatially and temporally variable habi- habitat on gene flow and to determine the extent to tat requirements, movement across unsuitable vege- which possible landscape resistance models explained tation types would need to occur to access newly the spatial pattern of GD between individuals. We cre- created habitat patches postfire. In our study, the ated a matrix of GD based on the same genetic distance eastern chestnut mouse was highly habitat specific measure used for the spatial autocorrelation analysis in occurrence across a landscape characterized by a described above (Smouse & Peakall 1999). We produced heterogeneous mosaic of vegetation types. Therefore, landscape resistance surfaces representing different we took two approaches to test the influence of veg- hypotheses about the difficulty in traversing different etation type on genetic connectivity: first we classi- habitat types using CIRCUITSCAPE version 3.5 (McRae & fied vegetation types as habitat vs. nonhabitat based Shah 2009). CIRCUITSCAPE uses circuit theory (McRae & on whether they were occupied or not and para- Beier 2007; McRae et al. 2008) to predict connectivity meterized the models according to the range of in heterogeneous landscapes and allows for multiple plausible conductivity values (with the conductance pathways of gene flow rather than a single corridor value of habitat ranging from equal to nonhabitat
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 174 Appendix 3: Manuscript – Landscape genetics
1272 F. PEREOGLOU ET AL.
(A) (B)
(C) (D)
Fig. 2 Habitat quality relating to movement across a null landscape (A), burned areas (B), selected vegetation types (C) and occupied habitat (D). Grey scale based on cell conductance values listed in Table 1. Darker colour indicates a lower resistance value assigned.
up to 1000 times greater than nonhabitat); and randomizations. To infer an effect of a landscape resis- second, we considered conductivity was proportional tance scenario on dispersal, we expected (i) and (ii) to to habitat quality as assessed by the percentage of be significant, and we expected (iii) to be nonsignificant sites of each vegetation type that were occupied if that scenario ‘correctly’ explained population connec- (Table 1). tivity in our study population (Cushman et al. 2006). As an alternative statistical test of the effect of land- We proposed nine alternative landscape models scape resistance after accounting for the ‘null’ model, (Table 1) that we tested against the ‘null landscape’ iso- we used multiple regression on distance matrices lation-by-distance hypothesis. We used Mantel and par- (MRDM; Lichstein 2007). tial Mantel correlations as the test statistics for causal modelling, including (i) simple Mantel tests between Results genetic distance and landscape resistances; (ii) partial Mantel tests between genetic distance and landscape Population genetic summary statistics resistances, partialling out the effects of the null model; and (iii) for those landscape resistance scenarios for Genotype proportions deviated from H–W expectations which (ii) was significant, partial Mantel tests between in 2008–2009 at locus 7D12 in site 7-P and in 2009–2010 genetic distance and the null model, partialling out at locus 9A8 in site 7-I, locus 10G6 in site 7-46 and locus the effects of the landscape resistance models. We 1A7 in site 7-95. Locus 6D4 showed consistent signifi- conducted Mantel and partial Mantel tests using the cant departures from HWE (in sites 7-44B, 7-2 and 7-H ECODIST package (Goslee & Urban 2007) in R version in 2008–2009 and at 7-44B, 7-46 and 7-M in 2009–2010) 2.7 (R Development Core Team 2008) with 10 000 and was removed from subsequent analyses. With the
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 175 Appendix 3: Manuscript – Landscape genetics
LANDSCAPE GENETICS OF A POSTFIRE SPECIALIST 1273
Table 1 Raster cell values used to create resistance landscapes
Model Hypothesis Cell conductance values
Null All habitat types exert the same level of All cells = 1 permeability to gene flow Firemodel1 Areas burnt in the 2003/2004 fire are twice as Burnt areas = 2 permeable to gene flow as unburned habitat All other cells = 1 Firemodel2a Areas burnt in the 2003/2004 fire are 10 times as Burnt areas = 10 permeable to gene flow as unburned habitat All other cells = 1 Firemodel2b Areas burnt in the 2003/2004 fire are 100 times as Burnt areas = 100 permeable to gene flow as unburned habitat All other cells = 1 Firemodel2c Areas burnt in the 2003/2004 fire are 1000 times as Burnt areas = 1000 permeable to gene flow as unburned habitat All other cells = 1 Vegmodel3† Vegetation types where mice have been captured Cells in heathland, woodland and sedgeland = 2 are twice as likely to facilitate gene flow Cells in other vegetation types = 1 Vegmodel4a† Vegetation types where mice have been captured Cells in heathland, woodland and sedgeland = 10 are 10 times as likely to facilitate gene flow Cells in other vegetation types = 1 Vegmodel4b† Vegetation types where mice have been captured Cells in heathland, woodland and sedgeland = 100 are 100 times as likely to facilitate gene flow Cells in other vegetation types = 1 Vegmodel4c† Vegetation types where mice have been captured Cells in heathland, woodland and sedgeland = 1000 are 1000 times as likely to facilitate gene flow Cells in other vegetation types = 1 Vegmodel6† Permeability to gene flow is associated with Cells in heathland, woodland and sedgeland = percentage of habitat quality surveyed sites of that vegetation type that were occupied (1–100) Cells in other vegetation types = 1
Water was assigned a value of zero. † Cell values were estimated from unpublished annual monitoring data (refer to Table S1, Supporting information).
Table 2 Population genetic parameters for the eastern chestnut mouse
2008–2009 2009–2010
Site nNa HO HE FIS nNa HO HE FIS
7-44 8 2.9 0.2 0.597 0.093 0.589 0.028 0.005 0.134 0 7-46 38 3.8 0.3 0.610 0.069 0.567 0.058 0.070 0.037 24 3.2 0.2 0.585 0.048 0.551 0.042 0.075 0.055 7-95 2 11 3.0 0.2 0.524 0.068 0.500 0.044 0.029 0.073 7-2 23 3.1 0.3 0.528 0.078 0.471 0.065 0.112 0.037 18 3.1 0.2 0.530 0.069 0.526 0.052 0.002 0.069 7-44B 21 3.3 0.2 0.579 0.038 0.556 0.039 0.048 0.031 17 3.3 0.2 0.572 0.024 0.569 0.028 0.021 0.055 7-H 23 2.9 0.3 0.499 0.058 0.462 0.055 0.093 0.040 21 2.9 0.2 0.415 0.055 0.434 0.055 0.032 0.055 7-I 19 3.3 0.2 0.567 0.045 0.569 0.020 0.008 0.066 14 3.1 0.2 0.492 0.067 0.473 0.060 0.018 0.093 7-M 3 13 2.9 0.2 0.632 0.028 0.519 0.038 0.246 0.059 7-P 12 3.0 0.2 0.652 0.067 0.555 0.039 0.172 0.093 8 2.8 0.2 0.621 0.062 0.520 0.048 0.203 0.077 Mean 3.2 0.1 0.576 0.025 0.538 0.018 0.069 0.027 3.0 0.1 0.547 0.020 0.511 0.016 0.070 0.025
The number of samples (n), the number of alleles (Na), observed heterozygosity (HO), expected heterozygosity (HE) and FIS for each site averaged across loci standard error. exception of locus 6D4, the departures from H–W Population subdivision and spatial genetic structure expectations in our data set were not beyond the false error rate for multiple testing (i.e. seven of 130 tests We found significant genetic differentiation among sites – = = were likely to be significant) nor were they consistent across the study area (2008 2009 FST 0.161, P 0.001; – = = across sites or years. Overall, we identified between 2009 2010 FST 0.195, P 0.001; Table S3, Supporting three and six alleles per locus. Genetic diversity across information) and an effect of isolation by distance. sites ranged from 0.499 to 0.652 in 2008 to 2009 and Euclidean distance (log-scale geographic distance) was – 0.415 to 0.632 in 2009 to 2010, with observed heterozy- a significant predictor of FST between sites (2008 2009 gosity averaged across sites greater than expected in Mantel r = 0.703, P = 0.007; 2009–2010 Mantel r = 0.377, both survey years (Table 2). P = 0.044 r = 0.377, P = 0.044; Fig. 3).
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 176 Appendix 3: Manuscript – Landscape genetics
1274 F. PEREOGLOU ET AL.
0.70 multilocus genotypes were less similar than those
0.60 assigned to distance classes at random. At a fine scale, the within-site distribution of genotypes was significant 0.50 and positively autocorrelated up to 200–250 m (Fig. 4C, st 0.40 D). /1-F
st 0.30 We did not identify discrete, spatially segregated, F
0.20 genetic populations above the scale of sampling grids using the clustering methods. Structure did identify the 0.10 presence of three groups in both years. However, it 0.00 > 2.5 3.0 3.5 4.0 could not assign individuals to a population with 90% Log geographic distance (m) confidence (Table S5, Supporting information). Further, these groups did not correspond to spatially discrete Fig. 3 Relationship between genetic distance (pairwise site FST) clusters of individuals, and we found that samples from and geographic distance. Solid symbol is 2008–2009 data; a particular site included a mixture of individuals from hollow symbol is 2009–2010 data. different structure groups (Appendix S2 and Figure S1, Supporting information). GENELAND identified seven At a broad scale, we detected positive and significant groups in 2008–2009 and eight in 2009–2010. Individuals spatial autocorrelation in distance classes up to 2 km were assigned to clusters that reflected population sub- – (Fig. 4A, B). At distances exceeding 3 3.5 km, pairwise division based on pairwise FST, suggesting field sites
0.200 Fig. 4 Correlograms indicating spatial auto- (A) 0.150 correlation of pairwise multilocus genetic – 0.100 distances over 5 km (A) 2008 2009, (B) 2009 –2010; and over 250 m (C) 2008–2009, (D)
r 0.050 2009–2010. Broken lines represent the 95% 0.000 CI for the null hypothesis of a random –0.050 distribution of genotypes. Error bars repre- –0.100 sent the 95% CI around r for a specific 500 1000 1500 2000 2500 3000 3500 4000 4500 5000 distance class.
0.200 0.150 (B) 0.100
r 0.050 0.000 –0.050 –0.100 500 1000 1500 2000 2500 3000 3500 4000 4500 5000
0.250 0.200 (C) 0.150 0.100 r 0.050 0.000 –0.050 –0.100 50 100 150 200 250
0.250 0.200 (D) 0.150 0.100 r 0.050 0.000 –0.050 –0.100 50 100 150 200 250 Distance class
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 177 Appendix 3: Manuscript – Landscape genetics
LANDSCAPE GENETICS OF A POSTFIRE SPECIALIST 1275 comprised distinct demographic units (Appendix S2 Landscape resistance and Table S4, Supporting information). However, GENE- LAND could not assign individuals to a population with 1 Simple Mantel tests between genetic distance and >90% confidence (Table S5, Supporting information). landscape resistances. All of the simple Mantel tests Broadly, GENELAND results matched with pairwise FST were significant (Table 3). Both the null model and between sites, but did not add any higher-level spatial the alternate resistance landscapes explained variation groupings of individuals. in genetic distance between individuals. However, Analyses in GESTE v2 identified a mean population- when the models were ranked based on Mantel r, only specific FST estimate of 0.105 (SD of 0.044 across sites; Firemodel1 performed better than the null model. Figure S2, Supporting information). The constant-only 2 Partial Mantel tests between genetic distance and model was best supported (posterior probabil- landscape resistances, partialling out the effects of the ity = 0.727). This suggests that none of the modelled null model. We found significant effects of the fire variables representing vegetation heterogeneity or models 1, 2a, 2b and 2c (conductance across burned spatial location affected population-specific FST. habitat was twice, 10, 100 and 1000 times greater than unburned habitat, respectively; Table 1) after account- ing for variation in genetic distance explained by the Sex-biased dispersal null landscape. After we accounted for the null We found some evidence that spatially proximal model, the partial correlations of the vegetation mod- females were more genetically similar than males in els (with increases in relative conductivity of pre- one of our analyses in on year of sampling; however ferred habitat to other vegetation types and with this pattern was inconsistent. We did not find strong conductance parameterised to reflect the distribution evidence for heterogeneity between correlograms for of capture records and habitat preferences) with males and females. In 2008–2009, female r-values were genetic distance were nonsignificant (Table 3). significantly greater than males in the 0- to 50-m dis- 3 Partial Mantel tests between genetic distance and the = tance class according to the T2 test (T2 4.182, null model, partialling out the effects of the landscape P = 0.033) but not in the 0- to 500-m distance class resistance models. Of the landscape resistance scenar- = = (T2 0.001, P 0.977). However, the bootstrap 95% ios that explained a significant component of the varia- CIs of the r values, a more conservative test (Banks & tion in genetic distance between individuals when the Peakall 2012) for males and females overlapped null model was partialled out (the four fire models: (Table S6, Supporting information). In 2009–2010, Table 1), the null model was nonsignificant after Fire- female r-values were not significantly greater than model1 was accounted for (Table 3). This model fea- = males in the 0- to 50-m distance class (T2 0.087, tured a conductance of recently burned habitat twice P = 0.751), but they were in the 0- to 500-m distance that of unburned habitat. The null model explained a = = class (T2 14.258, P 0.001). This was the only spatial significant component of the variation in genetic dis- autocorrelation value test of sex-biased dispersal to be tance even after the remaining fire models (Firemod- supported by the conservative bootstrap 95% CI over- el2a–c, featuring relatively higher conductance values lap test (Table S6, Supporting information). There were of burned habitat: Table 1) were fitted (Table 3). This no consistent biological patterns in the other distance suggests that the latter models potentially overesti- class comparisons in either survey year (Table S6, Sup- mate the relative conductivity of burnt habitat. porting information). None of the sex-biased dispersal tests in FSTAT provided evidence for female philopatry and male dispersal, and none of the test statistics were Discussion significant in either survey year (Table S7, Supporting information). The test statistics were conflicting in Our study provides support for a ‘generalist’ dispersal regard to expectations of male-biased dispersal (e.g. strategy, which would be advantageous for a species
FST, FIS and HS; Table S7, Supporting information). that prefers ephemeral habitat, conditional on spatial The lack of apparent sex-biased dispersal may be and temporal variability. We found significant genetic attributed to small sample sizes used in these analy- differentiation between sites that followed an isolation- ses. However, given that sex-biased dispersal can be by-distance pattern and significant spatial autocorrela- detected with similar sample sizes in species where tion between genotypes, indicating dispersal was this is a prevailing life-history strategy, such as restricted relative to the overall scale of the study. How- Antechinus species (Banks & Peakall 2012), our results ever, we did not find strong evidence for dispersal to are likely due to biological reasons discussed in the be male-biased and limited to only one sex. Our data next section. suggested no effect of heterogeneity in heterogeneous
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 178 Appendix 3: Manuscript – Landscape genetics
1276 F. PEREOGLOU ET AL.
Table 3 Results of Mantel and partial Mantel tests of landscape resistance on gene flow
Mantel test MRDM
Matrix 1 Matrix 2 Partial rM P-value P-value
(1) Simple Mantel tests between genetic distance and landscape resistances GD Null 0.215 <0.001 <0.001 GD Firemodel1 0.229 <0.001 <0.001 GD Firemodel2a 0.196 <0.001 <0.001 GD Firemodel2b 0.122 0.009 0.006 GD Firemodel2c 0.110 0.012 0.012 GD Vegmodel3 0.215 <0.001 <0.001 GD Vegmodel4a 0.194 <0.001 <0.001 GD Vegmodel4b 0.137 <0.001 <0.001 GD Vegmodel4c 0.117 <0.001 <0.001 GD Vegmodel6 0.157 <0.001 <0.001 (2) Partial Mantel tests between genetic distance and landscape resistances, partialling out the effects of the null model GD Firemodel1 Null 0.085 0.028 0.026 GD Firemodel2a Null 0.084 0.032 0.028 GD Firemodel2b Null 0.084 0.033 0.034 GD Firemodel2c Null 0.016 0.524 0.511 GD Vegmodel3 Null 0.038 0.180 0.174 GD Vegmodel4a Null 0.050 0.061 0.068 GD Vegmodel4b Null 0.054 0.053 0.055 GD Vegmodel4c Null 0.042 0.112 0.109 GD Vegmodel6 Null 0.033 0.322 0.316 (3) Partial Mantel tests between genetic distance and the null model, partialling out the effects of the null landscape resistance models GD Firemodel1 Null 0.029 0.405 0.308 GD Firemodel2a Null 0.125 <0.001 <0.001 GD Firemodel2b Null 0.197 <0.001 <0.001 GD Firemodel2c Null 0.204 <0.001 <0.001 GD Vegmodel3 Null 0.011 0.614 0.605 GD Vegmodel4a Null 0.098 <0.001 <0.001 GD Vegmodel4b Null 0.175 <0.001 <0.001 GD Vegmodel4c Null 0.19 <0.001 <0.001 GD Vegmodel6 Null 0.154 <0.001 <0.001
GD = pairwise multilocus genetic distance. Data for both survey years are combined. Significance levels from multiple regression on distance matrices (MRDM) are included for comparison.
native vegetation on genetic structure, despite strong of this system. The population-specific FST analysis in habitat selection for specific vegetation types, and there GESTE provided no evidence for genetic structure being was evidence for a positive effect of fire in promoting influenced by the distribution of preferred and nonpre- connectivity. ferred habitat (vegetation heterogeneity). Consistent with this finding, landscape genetics analyses of the relationships between pairwise individual GD and a set Landscape connectivity and gene flow of landscape resistance scenarios did not support an For a pyric specialist species with preference for early effect of vegetation-mediated habitat suitability on spa- successional stage vegetation, we predicted fire would tial genotypic patterns. This suggests that occupancy promote gene flow and enhance connectivity because and abundance are not appropriate surrogates for land- following burned vegetation would potentially lead scape permeability to movement in disturbance-adapted dispersers to unoccupied suitable habitat. While we species (Spear et al. 2010). However, these analyses did identified significant genetic differentiation between support an increased relative permeability of burnt populations, we found no evidence for discrete structure landscapes over unburnt landscapes, suggesting that fire above the level of sampling sites (i.e. broad spatial pop- promotes connectivity in this species. ulation groupings or clusters), suggesting that a model The support for increased permeability of burnt rela- of isolation by distance is an appropriate interpretation tive to unburnt habitat came from simple Mantel tests
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 179 Appendix 3: Manuscript – Landscape genetics
LANDSCAPE GENETICS OF A POSTFIRE SPECIALIST 1277 of interindividual genetic versus landscape resistance A postfire-collapsed shrub layer may even provide distances, in which Firemodel 1 (Table 1), where the cover and protection facilitating movement (Spear et al. conductance of recently burned habitat was twice that 2005). Spear & Storfer (2010) found gene flow was of unburned habitat, had the strongest simple Mantel maintained across burned areas and suggested move- correlation (Table 3). Although our Mantel correlations ment was facilitated by biological legacies (sensu Frank- were not high (maximum r = 0.229), we expected that lin et al. 2000) in the form of downed trees and debris. geographic correlations with individual multilocus dis- For dispersing individuals of the eastern chestnut tances to be ‘messier’ than population-based distances mouse, a fire boundary of unburned habitat or late suc- as the latter do not include intrapopulation (interindi- cessional stage vegetation might be a greater deterrent vidual) variation. Further, partial Mantel tests and to movement than the burnt landscape (i.e. habitat cues MRDM identified this model as being significantly asso- can assist directional movement; Bowler & Benton ciated with genetic distance after accounting for a ‘null’ 2005). Under this scenario, the underlying mechanism landscape model. This model that satisfied the classic for persistence of postfire specialist fauna may be the causal modelling criteria for acceptance (Cushman et al. disperser’s ability to follow fire in an attempt locate 2006) was Firemodel1 (Table 1), in which the conduc- suitable habitat. tance of recently burned habitat was twice that of Second, an alternative explanation is that the resis- unburned habitat. Landscape resistance scenarios in tance landscape for dispersal by the eastern chestnut which recently burned habitat had a conductance 10-, mouse is not static and that dispersal is increased dra- 100- or 1000-fold greater than unburned habitat were matically in burnt areas in association with colonization significant after accounting for the null model, but the immediately after the fire and then returns to levels null model was also significant after these models were described under a ‘null landscape’ model. Genetic sig- accounted for. This suggests that these latter models nals of recent (but not current) dispersal scenarios can overstate the relative conductance value of recently persist for a number of generations (Landguth et al. burned habitat and that Firemodel1 is a reasonably 2012). Under this scenario, fire causes substantial accurate representation of the resistance landscape for mortality but also results in rapid colonization of newly the studied population. While this finding was consis- created early successional habitat by residual survivors tent with our prediction for this species, and with other and thus increased gene flow over the immediate short- lines of evidence (the observed heterozygosity excess in term postfire. Once the empty territories have been both years is consistent with increased dispersal and colonized, there is no longer any reason for higher dis- admixture of populations that were likely to have been persal across burnt areas. Our data do not distinguish considerably smaller and more isolated before the 2003 these two explanations. fire), we do interpret the partial Mantel and MRDM For species with spatially and temporally variable analyses with some caution, as recent simulation work habitat requirements, we predicted natural habitat has shown that these tests can, under some circum- heterogeneity would not restrict gene flow because dis- stances, have unacceptable error rates for the signifi- persal would still need to occur through landscapes that cance of spatially structured explanatory variables do not meet specific habitat requirements. A heteroge- (Jaquiery et al. 2011; Guillot & Rousset 2012). Statistical neous cover of native vegetation types is unlikely to caveats aside, we discuss the implications of fire for restrict gene flow because dispersal is still possible connectivity in postfire specialists below. through landscapes that are considered inappropriate Potentially, fire promotes connectivity for early suc- when conditional on perceived habitat preferences. cessional specialist fauna in two ways. First, fire Although the eastern chestnut mouse has specific habi- improves habitat suitability for such species, and habi- tat requirements including early successional stage tat suitability improves landscape conductance to gene heathland vegetation, we found genetic connectivity flow. Where long-term fire suppression occurred in a across the study area was not limited by vegetation previously fire-maintained ecosystem, Templeton et al. type and that other native vegetation communities (2001) observed a significant increase in dispersal and interspersed throughout heathland did not restrict colonization rate of new patches by collared lizards movement. While barriers to dispersal may funnel after re-initiation of burning. The increase in dispersal movement in one direction (Gustafson & Gardner 1996), was attributed not to the fire itself but to the change in dispersal through unfavourable habitat has been structure of the intervening habitat matrix and opening recorded for many species (Ruscoe et al. 1998; Mossman of the understorey (Templeton et al. 2001). For example, & Waser 2001; Selonen & Hanski 2004; Gauffre et al. burned areas may not be an impediment to movement 2008; Macqueen et al. 2008). Selectivity for preferred as long as some structure, native plant regene- habitat during dispersal has also been recorded for ration or patches of remnant vegetation are present. other species (Lorenz & Barrett 1990; Baur & Baur 1992;
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 180 Appendix 3: Manuscript – Landscape genetics
1278 F. PEREOGLOU ET AL.
Schultz 1998; Lindenmayer & Peakall 2000; Banks et al. However, clear genetic signals of sex-biased dispersal 2005), indicating the effect (or not) of habitat heteroge- were detected using similar sample sizes for two other neity on gene flow is likely to be strongly species species that also occur in this landscape (or are closely specific and/or site specific. A species’ capacity for related to species in this landscape), the agile antechi- movement through heterogeneous landscapes is likely nus (Antechinus agilis) and the bush rat (Rattus fuscipes; to depend on whether heterogeneity is provided by Peakall et al. 2003; Banks et al. 2005; Banks & Peakall indigenous vegetation of varying composition (as dem- 2012). These species are not postfire specialists. So onstrated in our study) or by small, remnant patches of whatever the absolute level of dispersal by male and native vegetation interspersed with anthropogenic mod- female eastern chestnut mice, the evidence suggests that ified landscapes (i.e. agricultural land, non-native forest the relative strength of sex bias in dispersal is weaker plantations, built environments, etc.). in the eastern chestnut mouse than in other small mam- mals studied in this region. The lack of a signal of sex- biased dispersal (a consistent significant difference in Population genetic structure and dispersal behaviour pairwise spatial autocorrelation of male and female ani- A generalist dispersal strategy (i.e. one where move- mals) suggests an atypical mammalian structure exists ment is not restricted to particular vegetation types or (Gardner-Santana et al. 2009). This is because male- landscape attributes) is advantageous for a species that biased dispersal in mammals is considered an effective prefers ephemeral habitats that are less predictable and mechanism for inbreeding avoidance (Greenwood 1980; more variable in temporal and spatial dispersion of Bowler & Benton 2005). Eastern chestnut mouse resources (Lee et al. 1981). For small mammals in rela- dispersal behaviour may have evolved as a response to tively stable environments, dispersal is often restricted acquisition of suitable habitat rather than inbreeding (Peakall et al. 2003; Kraaijeveld-Smit et al. 2007; Mac- avoidance, and therefore, it is an advantageous strategy queen et al. 2008; Gardner-Santana et al. 2009). Tempo- for both sexes to be capable of dispersal. An interesting ral instability of habitat quality may enhance effective area for future research would be to investigate whether dispersal and/or increase dispersal propensity (Gauffre patterns of sex-biased dispersal differ between coloniza- et al. 2008). For example, a dispersal strategy with tion and migration (between existing populations) movement restricted to neighbouring habitat patches dispersal events, a point raised by Wade & McCauley would be ineffective for the eastern chestnut mouse to (1988). There is evidence that the tendency of one sex take advantage of recently burned heathland and access or another towards dispersal may be unpredictable and early successional stage vegetation. While there is not a dependent on habitat quality or landscape pattern clear relationship between dispersal distance and the (Banks et al. 2005). For example, in the Australian bush scale of genetic spatial autocorrelation, the scale of posi- rat Rattus fuscipes, sex-biased dispersal was absent in tive autocorrelation that we identified (2–3 km; Fig. 4) populations studied in south-eastern Australian forest is similar to the movement patterns of other pseudomy- (Kraaijeveld-Smit et al. 2007), but distinct female philop- ine mice. The smoky mouse Pseudomys fumeus and east- atry was found in fragmented rainforest in Queensland ern pebble-mound mouse P. patrius have both been (Macqueen et al. 2008). Further research into eastern observed to make minimum movements >600 m and chestnut mouse dispersal behaviour should be repli- 1 km, respectively, in a single night (Ford 2003, 2008). cated over multiple study areas before we are able to The sandy inland mouse P. hermannsburgensis is capable confirm the mechanisms postulated above. of dispersing very large distances (>16 km; Dickman et al. 1995), suggesting the eastern chestnut mouse may Conclusion be physiologically capable of much longer dispersal subject to landscape constraints. Landscape resistance was a better predictor of genetic Given that colonization is limited by the dispersal- structure in the eastern chestnut mouse population than limited sex, a generalist dispersal strategy that includes isolation by distance alone. Genetic connectivity was an absence of sex-biased dispersal would be beneficial provided by burnt habitat, indicating fire may play a for a postfire specialist species to access available high- positive role in structuring populations of some species. quality but unoccupied habitat. Our results show no Dispersal was possible through unsuitable habitat, and obvious evidence for sex bias in dispersal by the eastern any effect of landscape heterogeneity is likely to depend chestnut mouse. This indicates that both sexes disperse on whether heterogeneity is provided by indigenous or that our data set was not large enough to discern the vegetation or by anthropogenic modified landscapes. philopatric sex. Whether a low, or temporally inconsis- We postulate that a generalist dispersal strategy – tent, level of sex-biased dispersal occurs and was not where movement is not restricted to particular vegeta- detectable in our genetic data is difficult to determine. tion types, landscape attributes or sex biased – is
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 181 Appendix 3: Manuscript – Landscape genetics
LANDSCAPE GENETICS OF A POSTFIRE SPECIALIST 1279 advantageous for a species that prefers ephemeral habi- Cockburn A, Scott MP, Scotts DJ (1985) Inbreeding avoidance tat conditional on spatial and temporal variability. and male-biased natal dispersal in Antechinus spp. (Marsu- pialia: Dasyuridae). Animal Behaviour, 33, 908–915. Cushman SA, McKelvey KS, Hayden J, Schwartz MK (2006) Acknowledgements Gene flow in complex landscapes: testing multiple hypothe- ses with causal modeling. The American Naturalist, 168, 486– We most gratefully acknowledge the Wreck Bay Aboriginal 499. Community and the Wandandian People of the Dharawal and Dickman CR, Predavec M, Downey FJ (1995) Long-range move- Dhurga language groups on whose lands this study was con- ments of small mammals in arid Australia: implications for ducted; Christine Hayes for preliminary experimental laboratory land management. Journal of Arid Environments, 31, 441–452. analysis and assistance; Wendy Hartman, Alfred Wellingto, ESRI (2009) ArcMap 9.2. Environmental Systems Resource Insti- Tony Carter, Geoff Kay, Steve Phillips, Marama Hopkins, Corey tute, California. Green, Nick Dexter, Karen Viggers, Sally Townley, Les Mitchell Excoffier L, Foll M, Petit RJ (2009) Genetic Consequences of and Martin Westgate for field assistance and advice; and Rod Range Expansions. Annual Review of Ecology, Evolution, and Peakall for laboratory support and comments that helped Systematics, 40, 481–501. improve this manuscript. Funding was provided by the Austra- Fahrig L (2007) Non-optimal animal movement in human- lian Academy of Science, WV Scott Charitable Trust, Norman altered landscapes. Functional Ecology, 21, 1003–1015. Wettenhall Foundation and was conducted under ANU Animal Fahrig L, Paloheimo J (1988) Determinants of local population Experimentation Ethics Committee protocol C.RE.49.08, NSW size in patchy habitats. Theoretical Population Biology, 34, National Parks and Wildlife Service Scientific Licence S12690 194–213. and Booderee National Park Permit BDR08/00009. The com- Flannigan MD, Krawchuk MA, de Groot WJ, Wotton BM, ments of four anonymous reviewers and subject editor Lisette Gowman LM (2009) Implications of changing climate for glo- Waits substantially improved earlier versions of this manuscript. bal wildland fire. International Journal of Wildland Fire, 18, 483 –507. References Foll M, Gaggiotti O (2006) Identifying the environmental fac- tors that determine the genetic structure of Populations. Baguette M, Van Dyck H (2007) Landscape connectivity and Genetics, 174, 875–891. animal behavior: functional grain as a key determinant for Ford FD (2003) Conilurine Rodent Evolution: The Role of Ecology dispersal. Landscape Ecology, 22, 1117–1129. in Modifying Evolutionary Consequences of Environmental Banks SC, Peakall R (2012) Genetic spatial autocorrelation can Change. James Cook University, Townsville, Australia. readily detect sex-biased dispersal. Molecular Ecology, 21, Ford FD (2008) Smokey mouse, Pseudomys fumeus. In: The Mam- 2092–2105. mals of Australia (eds Van Dyck S, Strahan R), pp. 629–630. Banks SC, Lindenmayer DB, Ward SJ, Taylor AC (2005) The Sydney, Australia, Reed New Holland. effects of habitat fragmentation via forestry plantation Fox BJ (1982) Fire and mammalian secondary succession in an establishment on spatial genotypic structure in the small Australian coastal heath. Ecology, 63, 1332–1341. marsupial carnivore, Antechinus agilis. Molecular Ecology, 14, Fox BJ (2008) Eastern chestnut mouse, Pseudomys gracilicauda- 1667–1680. tus. In: The Mammals of Australia (eds Van Dyck S, Strahan Banks SC, Dujardin M, McBurney L, Blair D, Barker M, R), pp. 634–635. Sydney, Australia, Reed New Holland. Lindenmayer DB (2011) Starting points for small mammal Fox BJ, Kemper CM (1982) Growth and development of Pseudo- population recovery after wildfire: recolonisation or residual mys gracilicaudatus (Rodentia: Muridae) in the laboratory. populations? Oikos, 120,26–37. Australian Journal of Zoology, 30, 175–185. Baur A, Baur B (1992) Effect of corridor width on animal dis- Fox BJ, Taylor JE, Thompson PT (2003) Experimental manip- persal: a simulation study. Global Ecology and Biogeography ulation of habitat structure: a retrogression of the small Letters, 2,52–56. mammal succession. Journal of Animal Ecology, 72, 927–940. Bond WJ, Keeley JE (2005) Fire as a global herbivore: the Franklin J, Lindenmayer D, MacMahon J (2000) Threads of ecology and evolution of flammable ecosystems. Trends in continuity: ecosystem disturbances, biological legacies and Ecology & Evolution, 20, 387–394. ecosystem recovery. Conservation Biology in Practice, 1,8–16. Bowler DE, Benton TG (2005) Causes and consequences of Gardner-Santana LC, Norris DE, Fornadel CM, Hinson ER, animal dispersal strategies: relating individual behaviour to Klein SL, Glass GE (2009) Commensal ecology, urban land- spatial dynamics. Biological Reviews, 80, 205–225. scapes, and their influence on the genetic characteristics of Bruford MW, Hanotte O, Brookfield JKY, Burke T (1988) city-dwelling Norway rats (Rattus norvegicus). Molecular Ecol- Multilocus and single-locus DNA fingerprinting. In: Molecu- ogy, 18, 2766–2778. lar Genetic Analysis of Populations: A Practical Approach (ed. Gauffre B, Estoup A, Bretagnolle V, Cosson JF (2008) Spatial Hoelzel AR), pp. 287–336. IRL Press, Oxford. genetic structure of a small rodent in a heterogeneous land- Catling PC, Newsome AE (1981) Responses of the Australian scape. Molecular Ecology, 17, 4619–4629. vertebrate fauna to fire: an evolutionary approach. In: Fire Goslee SC, Urban DL (2007) The ecodist package for dissimi- and the Australian Biota (eds Gill AM, Groves RH, Noble IR), larity-based analysis of ecological data. Journal of Statistical pp. 273–310. Canberra, Australian Academy of Science. Software, 22,1–19. Goudet J (2001) FSTAT, a program to estimate and test gene Clarke MF (2008) Catering for the needs of fauna in fire man- agement: science or just wishful thinking? Wildlife Research, diversities and fixation indices (version 2.9.3). Available from 35, 385–394. http://www.unil.ch/izea/softwares/fstat.html.
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 182 Appendix 3: Manuscript – Landscape genetics
1280 F. PEREOGLOU ET AL.
Goudet J, Perrin N, Waser P (2002) Tests for sex-biased dispersal Meek PD, Triggs B (1997) Extension to the range of the eastern using bi-parentally inherited genetic markers. Molecular Ecol- chestnut mouse, Pseudomys gracilicaudatus, to Jervis Bay, ogy, 11, 1103–1114. New South Wales. Australian Mammalogy, 20, 123–125. Greenwood PJ (1980) Mating systems, philopatry and dispersal Monamy V, Fox BJ (2000) Small mammal succession is deter- in birds and mammals. Animal Behaviour, 28, 1140–1162. mined by vegetation density rather than time elapsed since Gros A, Hovestadt T, Poethke HJ (2008) Evolution of sex- disturbance. Austral Ecology, 25, 580–587. biased dispersal: the role of sex-specific dispersal costs, Moro D, Spencer PBS (2003) Microsatellite primers for the wes- demographic stochasticity, and inbreeding. Ecological Model- tern pebble-mound mouse (Pseudomys chapmani) that show ling, 219, 226–233. cross amplification for other species of Australian rodent. Guillot G, Rousset F (2012) On the use of the simple and Molecular Ecology Notes, 3, 259–261. partial Mantel tests in presence of spatial auto-correlation. Morris DW (1992) Scales and costs of habitat selection in heter- arXiv:1112.0651v1. ogeneous landscapes. Evolutionary Ecology, 6, 412–432. Guillot G, Mortier F, Estoup A (2005) Geneland: a computer pack- Mossman CA, Waser PM (2001) Effects of habitat fragmenta- age for landscape genetics. Molecular Ecology Notes, 5,712–715. tion on population genetic structure in the white-footed Gustafson EJ, Gardner RH (1996) The effect of landscape heter- mouse (Peromyscus leucopus). Canadian Journal of Zoology- ogeneity on the probability of patch colonization. Ecology, 77, Revue Canadienne De Zoologie, 79, 285–295. 94–107. Murphy MA, Dezzani R, Pilliod DS, Storfer A (2010) Land- Jaquiery J, Broquet T, Hirzel AH, Yearsley J, Perrin N (2011) scape genetics of high mountain frog metapopulations. Inferring landscape effects on dispersal from genetic dis- Molecular Ecology, 19, 3634–3649. tances: how far can we go? Molecular Ecology, 20, 692–705. Park SDE (2001) Trypanotolerance in West African Cattle and the Kraaijeveld-Smit FJL, Lindenmayer DB, Taylor AC, MacGregor Population Genetic Effects of Selection. University of Dublin, C, Wertheim B (2007) Comparative genetic structure reflects Dublin, Ireland. underlying life histories of three sympatric small mammal Pavlacky DC, Goldizen AW, Prentis PJ, Nicholls JA, Lowe AJ species in continuous forest of south-eastern Australia. Oikos, (2009) A landscape genetics approach for quantifying the 116, 1819–1830. relative influence of historic and contemporary habitat heter- Landguth EL, Fedy BC, Oyler-McCance SJ et al. (2012) Effects ogeneity on the genetic connectivity of a rainforest bird. of sample size, number of markers, and allelic richness on Molecular Ecology, 18, 2945–2960. the detection of spatial genetic pattern. Molecular Ecology Peakall R, Smouse PE (2006) GenAlEx 6: genetic analysis in Resources, 12, 276–284. Excel. Population genetic software for teaching and research. Lee AK, Baverstock PR, Watts CHS (1981) Rodents—the late Molecular Ecology Notes, 6, 288–295. invaders. In: Ecological Biogeography of Australia (ed. Keast A), Peakall R, Ruibal M, Lindenmayer DB (2003) Spatial auto- pp. 1521–1553. Dr W Junk Publishers, The Hague, London. correlation analysis offers new insights into gene flow in Lichstein J (2007) Multiple regression on distance matrices: a the Australian bush rat, Rattus fuscipes. Evolution, 57, 1182– multivariate spatial analysis tool. Plant Ecology, 188, 117–131. 1195. Lindenmayer DB, Fischer J (2006) Habitat Fragmentation and Pereoglou F, Macgregor C, Banks SC, Ford FD, Wood JT, Landscape Change. Island Press, Washington DC. Lindenmayer DB (2011) Refuge site selection by the eastern Lindenmayer DB, Peakall R (2000) The Tumut experiment— chestnut mouse in recently burnt heath. Wildlife Research, 38, integrating demographic and genetic studies to unravel frag- 290–298. mentation effects: a case study of the native bush rat. In: Perez-Espona S, Perez-Barberia FJ, McLeod JE, Jiggins CD, Genetics, Demography and the Viability of Fragmented Popula- Gordon IJ, Pemberton JM (2008) Landscape features affect tions (eds Young A, Clarke G), pp. 173–201. Cambridge Uni- gene flow of Scottish Highland red deer (Cervus elaphus). versity Press, Cambridge. Molecular Ecology, 17, 981–996. Lorenz GC, Barrett GW (1990) Influence of simulated land- Pritchard JK, Stephens M, Donnelly P (2000) Inference of popu- scape corridors on house mouse (Mus musculus) dispersal. lation structure using multilocus genotype data. Genetics, American Midland Naturalist, 123, 348–356. 155, 945–959. Macqueen PE, Nicholls JA, Hazlitt SL, Goldizen AW (2008) R Development Core Team (2008) R: A Language and Environ- Gene flow among native bush rat, Rattus fuscipes (Rodentia: ment for Statistical Computing. R Foundation for Statistical Muridae), populations in the fragmented subtropical forests Computing, Vienna, Austria. of south-east Queensland. Austral Ecology, 33, 585–593. Ruscoe WA, Mather PB, Wilson J (1998) Genetic structure in Manel S, Schwartz MK, Luikart G, Taberlet P (2003) Landscape populations of Rattus sordidus in sugarcane–growing districts genetics: combining landscape ecology and population genet- of Queensland, Australia. Journal of Mammalogy, 79, 612–623. ics. Trends in Ecology and Evolution, 18, 189–197. Schrey AW, Fox AM, Mushinsky HR, McCoy ED (2010) Fire McRae BH, Beier P (2007) Circuit theory predicts gene flow in increases variance in genetic characteristics of Florida Sand plant and animal populations. Proceedings of the National Acad- Skink (Plestiodon reynoldsi) local populations. Molecular emy of Sciences of the United States of America, 104, 19885–19890. Ecology, 20,56–66. McRae BH, Shah VB (2009) Circuitscape User Guide. University Schrey AW, Ashton KG, Heath S, McCoy ED, Mushinsky HR of California, Santa Barbara. Available from http://www.cir- (2011) Fire Alters Patterns of Genetic Diversity Among 3 Liz- cuitscape.org. ard Species in Florida Scrub Habitat. Journal of Heredity, 102, McRae BH, Dickson BG, Keitt TH, Shah VB (2008) Using circuit 399–408. theory to model connectivity in ecology, evolution and con- Schuelke M (2000) An economic method for the fluorescent servation. Ecology, 89, 2712–2724. labeling of PCR fragments. Nature Biotechology, 18, 233–234.
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 183 Appendix 3: Manuscript – Landscape genetics
LANDSCAPE GENETICS OF A POSTFIRE SPECIALIST 1281
Schultz CB (1998) Dispersal behavior and its implications Williams AAJ, Karoly DJ, Tapper N (2001) The sensitivity of for reserve design in a rare oregon butterfly. Conservation Australian fire danger to climate change. Climatic Change, 49, Biology, 12, 284–292. 171–191. Scribner KT, Blanchong JA, Bruggeman DJ et al. (2005) Geo- Wright S (1943) Isolation by distance. Genetics, 28, 139–156. graphical genetics: conceptual foundations and empirical applications of spatial genetic data in wildlife management. 69 – Journal of Wildlife Management, , 1434 1453. F.P., S.B., J.W., F.F. and D.L. designed the research; F.P., C.M., Selonen V, Hanski IK (2004) Young flying squirrels (Pteromys J.W., D.L. and S.B. performed the research; F.P., volans) dispersing in fragmented forests. Behavioral Ecology, J.W. and S.B. analysed data; F.P., S.B. and D.L. wrote the 15, 564–571. manuscript; F.F., C.M. and J.W. provided editorial comment on Short Bull RA, Cushman SA, Mace R et al. (2011) Why replica- the manuscript. tion is important in landscape genetics: American black bear in the Rocky Mountains. Molecular Ecology, 20, 1092–1107. Smouse PE, Peakall R (1999) Spatial autocorrelation analysis of individual multiallele and multilocus genetic structure. Heredity, 82, 561–573. Data accessibility Smouse PE, Peakall R, Gonzales E (2008) A heterogeneity test for fine-scale genetic structure. Molecular Ecology, 17, 3389–3400. Appendix S2 (Supporting information) includes an Excel Spear SF, Storfer A (2008) Landscape genetic structure of file with separate worksheets containing the genotypes coastal tailed frogs (Ascaphus truei) in protected vs. managed and locations of individuals sampled in 2008/2009 and forests. Molecular Ecology, 17, 4642–4656. 2009/2010, as well as a combined genotypes sheet. Also Spear SF, Storfer A (2010) Anthropogenic and natural distur- included are the locations of each sampling site, the bance lead to differing patterns of gene flow in the Rocky recent fire history and vegetation characteristics of the Mountain tailed frog, Ascaphus montanus. Biological Conserva- tion, 143, 778–786. landscape around each site, and a sheet containing pair- Spear SF, Peterson CR, Matocq MD, Storfer A (2005) Landscape wise genetic distances between individuals, Euclidean genetics of the blotched tiger salamander (Ambystoma tigri- geographic distances and the resistance distances corre- num melanostictum). Molecular Ecology, 14, 2553–2564. sponding to each model of landscape resistance (Table 1). Spear SF, Balkenhol N, Fortin M-J, McRae BH, Scribner KIM (2010) Use of resistance surfaces for landscape genetic stud- ies: considerations for parameterization and analysis. Molecu- Supporting information – lar Ecology, 19, 3576 3591. Additional supporting information may be found in the online Steffen W, Burbidge AA, Hughes L et al. (2009) Australia’s Bio- version of this article. diversity and Climate Change. CSIRO Publishing, Melbourne. Storfer A, Murphy MA, Spear SF, Holderegger R, Waits LP Appendix S1 Methods for Structure and GENELAND analyses. (2010) Landscape genetics: where are we now? Molecular Ecology, 19, 3496–3514. Appendix S2 Results for Structure and GENELAND analyses. Taws N (1997) Vegetation Survey and Mapping of Jervis Bay Terri- Table S1 Number of sites surveyed, sites with eastern chestnut tory. Report, Environment Australia, Canberra. mouse captures, and occupancy by vegetation type. Taylor PD, Fahrig L, Henein K, Merriam G (1993) Connectivity is a vital element of landscape structure. Oikos, 68, 571–573. Table S2 PCR conditions and characteristics of microsatellites. Templeton AR, Robertson RJ, Brisson J, Strasburg J (2001) Dis- rupting evolutionary processes: the effect of habitat fragmen- Table S3 Pairwise site FST for the eastern chestnut mouse (FST – – tation on collared lizards in the Missouri Ozarks. Proceedings for 2008 2009 below the diagonal, 2009 2010 above). of the National Academy of Sciences of the United States of Amer- Table S4 Results of clustering analysis in GENELAND. ica, 98, 5426–5432. Templeton AR, Brazeal H, Neuwald JL (2011) The transition from Table S5 The number of individuals assigned to each popula- isolated patches to a metapopulation in the eastern collared liz- tion cluster in GENELAND. ard in response to prescribed fires. Ecology, 92, 1736–1747. Turner MG, Baker WL, Peterson CJ, Peet RK (1998) Factors Table S6 Results of sex-biased spatial autocorrelation analysis influencing succession: lessons from large, infrequent natural for both survey years. disturbances. Ecosystems, 1, 511–523. Wade MJ, McCauley DE (1988) Extinction and Recolonization: Table S7 Results of sex-biased dispersal test for both survey their Effects on the Genetic Differentiation of Local Popula- years using FSTAT. tions. Evolution, 42, 995–1005. Fig. S1 Number of population clusters identified using Struc- Wang Y-H, Yang K-C, Bridgman C, Lin L-K (2008) Habitat ture. suitability modeling to correlate gene flow with landscape – connectivity. Landscape Ecology, 23, 989 1000. Fig. S2 Estimates of population-specific FST (HPDIs) from Watts CHS, Kemper CM (1989) Muridae. In: Fauna of Australia, GESTE. The 95% highest posterior density intervals of GESTE Mammalia (eds Walton DW, Richardson BJ), pp. 939–956. population-specific FST estimates (x-axis) for the nine popula- AGPS, Canberra. tions with the largest sample sizes (y-axis).
© 2013 Blackwell Publishing Ltd
Pereoglou, F., Population processes in an early successional heathland 184
Appendix 4: Supporting information for Chapter 3
Supporting information for Chapter 3 Site occupancy
Pereoglou, F., Population processes in an early successional heathland 185 Appendix 4:Supporting information
Appendix 4. Supporting Information for Chapter 3 Site occupancy
Landscape, fire and habitat: which features of recently burned heathland influence site occupancy of an early successional specialist?
Felicia Pereoglou Christopher MacGregor Sam C. Banks Jeff Wood Fred Ford David B. Lindenmayer. Landscape Ecology, 31(2): 255-269
Appendix S1: Study species Eastern chestnut mouse (Pseudomys gracilicaudatus) The eastern chestnut mouse is an endemic Australian murid rodent. It has a disjunct distribution along the east coast of Australia and is absent throughout much of its range. The Jervis Bay region supports the most southerly contemporary population of the species (Meek and Triggs 1997). Usually reported at low density, the eastern chestnut mouse has been recorded in grassy open woodland to open dry sclerophyll forest, but is more often found in heathland (Fox 2008) where populations peak in abundance post-fire before declining over time as vegetation regenerates (Fox 1982). Threats to the viability of populations of this species’ are thought to include fire suppression, loss of heath habitat, and predation (Fox 2008). The eastern chestnut mouse is an annual summer breeder capable of producing multiple litters of one to five (usually three) offspring (Fox 2008). Early development is rapid which contrasts with a slow growth rate to reach maturity (Fox & Kemper 1982). Adults show significant sexual dimorphism (Fox & Kemper 1982) and attain average weights of 90 g for males and 75 g for females (Fox 2008). The species’ lifespan is less than two years in wild populations (Watts & Kemper 1989). Studies of the eastern chestnut mouse dietary requirements indicate it is a generalist herbivore, opportunistic in foraging and less specialised in food selection than any other species within the genus (Luo et al. 1994). The eastern chestnut mouse is terrestrial, mostly nocturnal, and individuals select diurnal refuges associated with the presence of specific structures including grass trees (Xanthorrhea spp.), buttongrass (Gymnoschoenus sp.), collapsed dead stems, and areas of tall, dense vegetation (Pereoglou et al. 2011). The eastern chestnut mouse is transient and non-gregarious in its use of shelter sites (Pereoglou et al. 2011). Little is known about social organisation and behaviour. Radio-tracking data suggests that adults of the eastern chestnut mouse are solitary and territorial, with females having a home range of 0.5 – 1 ha and up to 4 ha for males (Pereoglou et al. unpublished data).
Bush rat (Rattus fuscipes) The bush rat is a common and widespread terrestrial murid rodent that occurs throughout many parts of eastern Australia (Watts & Aslin 1981). The species can be found in many
Pereoglou, F., Population processes in an early successional heathland 186 Appendix 4:Supporting information different habitats and is widespread in eucalypt forest and areas where there is a dense understory (Breed and Ford 2007). Weights in excess of 200 g have been recorded for adult animals (Lunney 2008a). Bush rats are capable of breeding all year round under favourable (wet) conditions, but births predominantly occur in late spring and summer with litter size ranging from one to seven (average five) (Watts and Aslin 1981). Multiple litters can be produced in the breeding season, with animals born early attaining sexual maturity in the same season, and all animals capable of breeding in the following spring (Watts and Aslin 1981). Mortality occurs after breeding and the average lifespan of most animals is one year (Watts and Aslin 1981); few individuals (< 5%) survive to breed a second year (Robinson 1987). The bush rat has an omnivorous diet of fungi, seeds, fruit, plant tissue and arthropods (Warneke 1971). The species is nocturnal but not strictly territorial; home ranges can overlap (Lunney 2008a) with male territories being larger than those of females (Watts and Aslin 1981).
Swamp rat (R. lutreolus) The swamp rat is a native murid rodent distributed over coastal and sub-coastal south- eastern Australia (Lunney 2008b). The species occurs predominantly in heathland and sedgeland (Braithwaite 1982), and prefers areas of wet, dense cover (Lunney 2008b). Adult animals attain an average weight of approximately 200 g (Lunney 2008b). Breeding occurs in summer (Watts and Aslin 1981). Like its congener, the bush rat, the swamp rat is capable of breeding year around provided conditions are favourable. Females can produce multiple litters of three to five young in a single season, and offspring born early are able to breed in the season of their birth (Lunney 2008b). Longevity is also probably similar to the bush rat, being mostly annual. The swamp rat is a specialist herbivore whose main diet consists of grasses and sedges (Watts and Aslin 1981). The species shelters in burrows or builds above-ground nests in waterlogged areas (Lunney 2008b). The swamp rat is partially diurnal and moves through runways constructed in dense vegetation (Watts and Aslin 1981). Male home ranges are larger than non-overlapping female ranges, and individuals exhibit territorial social behaviour throughout much of the year (Lunney 2008b). The swamp rat is immediately and adversely affected by fire due to the species’ dependence on dense cover (Lunney 2008b) and relatively specialised diet (Braithwaite 1982).
Brown antechinus (Antechinus stuartii) The brown antechinus is a small dasyurid marsupial, widespread in forest and heathland throughout its coastal distribution from southern Queensland to southern New South Wales, Australia (Crowther 2002). The brown antechinus is mostly terrestrial although it also can be arboreal depending on the habitat. High density populations are found in areas with thick ground cover and abundant logs (Crowther and Braithwaite 2008). Adult males and females
Pereoglou, F., Population processes in an early successional heathland 187 Appendix 4:Supporting information attain average weights of 35 g and 20 g, respectively. Mating occurs in winter, and females typically have 8 – 10 pouch young. Multiple paternity is common (Kraaijeveld-Smit et al. 2002), with most litters being sired by two or three males. All males die at the end of a two- week mating season, although some females may survive to breed in a second year (Crowther and Braithwaite 2008). The brown antechinus is an opportunistic insectivore (Fox and Archer 1984). Ranging behaviour is probably similar to the agile antechinus (A. agilis), with males covering an area three times greater than that of females (Dickman 2008). Individuals nest communally, usually in hollow logs of tree trunks. Male-biased post-natal dispersal occurs in summer and individuals may continue to move between communal nests prior to the breeding season (Cockburn and Lazenby-Cohen 1992).
Pereoglou, F., Population processes in an early successional heathland 188 Appendix 4:Supporting information
Table S1: Ratio of eastern chestnut mouse captures to recaptures across vegetation type and survey years. The number of sites surveyed is provided in parentheses.
Vegetation type 2003 2003/04 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11 Total Forest 0 (18) 2:1 (24) 0 (34) 0 (37) 0 (24) 0 (23) 0 (25) 0 (24) 0 (11) 2 Rainforest 0 (8) 0 (8) 0 (15) 0 (9) 0 (8) 0 (8) 0 (8) 0 (8) 0 (3) 0 Woodland 0 (15) 0 (23) 0 (31) 0 (38) 0 (23) 2:1 (23) 5:2 (23) 4:3 (23) 0 (12) 11 Heathland 1:1 (13) 1:1 (26) 2:2 (35) 15:5 (43) 3:3 (26) 25:11 (26) 59:14 (26) 35:11 (25) 6:2 (13) 147 Shrubland 0 (16) 0 (20) 0 (29) 0 (31) 0 (20) 1:1 (20) 0 (20) 2:1 (19) 0 (9) 3 Sedgeland 0 (6) 1:1 (8) 0 (11) 0 (11) 0 (8) 0 (7) 5:2 (8) 2:1 (7) 0 (3) 8 Total 1:1 (76)A 4:3 (109) 2:2 (155)B 15:5 (169)B 3:3 (109) 28:13 (107)C 69:18 (110)B 43:16 (106)C 6:2 (51)D 171 ASite establishment was impeded by wildfire BSome sites were trapped more than once during the survey year CSome sites were unable to be surveyed due to flooding DSome sites were not surveyed due to logistical reasons
Table S2: Hierarchical generalised linear model parameters and results summary
Random Offset Fixed Explanatory Dependent Estimate S.E. t p Species associations Site Log effort Survey year Bush rat Eastern chestnut mouse -0.11 0.05 -2.06 0.04 Site Log effort Survey year Brown antechinus Eastern chestnut mouse -0.07 0.06 -1.20 0.23 Site Log effort Survey year Swamp rat Eastern chestnut mouse -0.02 0.09 -0.19 0.85 Site Log effort Survey year Bush rat+Brown antechinus Eastern chestnut mouse -0.10 0.53 -1.84 0.07 -0.04 0.06 -0.66 0.51 Site Log effort Survey year Bush rat+Swamp rat Eastern chestnut mouse -0.11 0.05 -2.06 0.04 -0.02 0.10 -0.23 0.82 Site Log effort Survey year Swamp rat+Brown antechinus Eastern chestnut mouse 0.01 0.10 0.10 0.92 -0.07 0.06 -1.19 0.23 Site Log effort Survey year Bush rat+Swamp rat+Brown antechinus Eastern chestnut mouse -0.10 0.05 -1.83 0.07 -0.01 0.10 -0.07 0.94
Pereoglou, F., Population processes in an early successional heathland 189 Appendix 4:Supporting information
-0.04 0.07 -0.62 0.53 Site Log effort Survey year Eastern chestnut mouse Bush rat -0.10 0.07 -1.31 0.19 Site Log effort Survey year Brown antechinus Bush rat 0.05 0.03 1.78 0.08 Site Log effort Survey year Swamp rat Bush rat -0.03 0.04 -0.77 0.44 Site Log effort Survey year Eastern chestnut mouse+Brown antechinus Bush rat -0.10 0.07 -1.33 0.19 0.05 0.03 1.79 0.07 Site Log effort Survey year Eastern chestnut mouse+Swamp rat Bush rat -0.10 0.07 -1.29 0.20 -0.03 0.04 -0.74 0.46 Site Log effort Survey year Swamp rat+Brown antechinus Bush rat -0.06 0.04 -1.47 0.14 0.06 0.03 2.23 0.03 Eastern chestnut mouse+Swamp Site Log effort Survey year rat+Brown antechinus Bush rat -0.10 0.07 -1.31 0.19 -0.06 0.04 -1.46 0.15 0.06 0.03 2.24 0.03 Site Log effort Survey year Bush rat Swamp rat 0.03 0.03 0.93 0.36 Site Log effort Survey year Browney year antechinus Swamp rat 0.16 0.04 4.00 <0.001 Site Log effort Survey year Eastern chestnut mouse Swamp rat 0.19 0.10 1.82 0.07 Site Log effort Survey year Bush rat+Brown antechinus Swamp rat -0.01 0.03 -0.39 0.70 0.17 0.04 3.84 <0.001 Site Log effort Survey year Bush rat+Eastern chestnut mouse Swamp rat 0.03 0.03 0.98 0.33 0.19 0.11 1.85 0.07 Site Log effort Survey year Eastern chestnut mouse+Brown antechinus Swamp rat 0.22 0.10 2.16 0.03 0.16 0.04 4.17 <0.001
Bush rat+Eastern chestnut mouse+Brown Site Log effort Survey year antechinus Swamp rat -0.01 0.03 -0.41 0.68 0.22 0.10 2.15 0.03
0.17 0.04 4.00 <0.001
Site Log effort Survey year Bush rat Brown antechinus 0.02 0.02 1.12 0.27 vey year Site Log effort Survey year Eastern chestnut mouse Brown antechinus 0.00 0.05 0.00 1.00
Site Log effort Survey year Swamp rat Brown antechinus 0.09 0.03 2.65 0.01 Site Log effort Survey year Bush rat+Eastern chestnut mouse Brown antechinus 0.02 0.02 1.12 0.26 0.00 0.05 -0.06 0.95 Site Log effort Survey year Bush rat+Swamp rat Brown antechinus 0.02 0.02 1.02 0.31 0.09 0.03 2.60 0.01 Site Log effort Survey year Swamp rat+Eastern chestnut mouse Brown antechinus 0.09 0.03 2.64 0.01
Pereoglou, F., Population processes in an early successional heathland 190 Appendix 4:Supporting information
0.00 0.05 -0.02 0.99 Bush rat+Swamp rat+Eastern chestnut Site Log effort Survey year mouse Brown antechinus 0.02 0.02 1.02 0.31 0.09 0.03 2.59 0.01 0.00 0.05 -0.06 0.95
Fire history Site Log effort Survey year Fires Eastern chestnut mouse 0.08 0.12 0.66 0.51 Site Log effort Survey year Fires Bush rat -0.10 0.08 -1.31 0.19 Site Log effort Survey year Fires Swamp rat -0.10 0.16 -0.63 0.53 Site Log effort Survey year Fires Brown antechinus 0.03 0.07 0.39 0.70 Site Log effort Survey year Burned 2003 Eastern chestnut mouse 0.86 0.69 1.25 0.21 Site Log effort Survey year Burned 2003 Bush rat -0.45 0.27 -1.66 0.10 Site Log effort Survey year Burned 2003 Swamp rat -1.21 0.44 -2.75 0.01 Site Log effort Survey year Burned 2003 Brown antechinus -0.79 0.30 -2.64 0.01
Site-level habitat quality Site Log effort Survey year Floristic richness Eastern chestnut mouse -0.03 0.03 -1.05 0.29 Site Log effort Survey year Floristic richness Bush rat -0.04 0.03 -1.28 0.20 Site Log effort Survey year Floristic richness Swamp rat -0.03 0.05 -0.61 0.54 Site Log effort Survey year Floristic richness Brown antechinus -0.02 0.03 -0.70 0.48 Site Log effort Survey year Cover Eastern chestnut mouse -0.01 0.01 -0.38 0.70 Site Log effort Survey year Cover Bush rat 0.00 0.01 -0.48 0.63 Site Log effort Survey year Cover Swamp rat 0.03 0.02 1.36 0.17 Site Log effort Survey year Cover Brown antechinus 0.00 0.01 0.33 0.74 Site Log effort Survey year Dead shrubs Eastern chestnut mouse 0.01 0.01 2.48 0.01 Site Log effort Survey year Dead shrubs Bush rat 0.00 0.00 0.28 0.78 Site Log effort Survey year Dead shrubs Swamp rat -0.02 0.01 -1.76 0.08 Site Log effort Survey year Dead shrubs Brown antechinus 0.00 0.00 0.12 0.90 Site Log effort Survey year Rock Eastern chestnut mouse 0.02 0.03 0.79 0.43 Site Log effort Survey year Rock Bush rat -0.03 0.05 -0.77 0.44 16.5 Site Log effort Survey year Rock Swamp rat -5.86 7 -0.35 0.72 Site Log effort Survey year Rock Brown antechinus -0.02 0.03 -0.51 0.61
Pereoglou, F., Population processes in an early successional heathland 191 Appendix 4:Supporting information
Site Log effort Survey year Bare Eastern chestnut mouse 0.00 0.02 0.07 0.95 Site Log effort Surveyeffort year Bare Bush rat 0.00 0.01 0.45 0.65 Site Log effort Survey year Bare Swamp rat -0.03 0.02 -1.87 0.06 Site Log effort Survey year Bare Brown antechinus -0.02 0.01 -2.10 0.04 Site Log effort Survey year Regenerating shrubs Eastern chestnut mouse 0.01 0.01 1.17 0.24 Site Log effort Survey year Regenerating shrubs Bush rat -0.01 0.01 -1.53 0.13 Site Log effort Survey year Regenerating shrubs Swamp rat 0.00 0.01 -0.17 0.86 Site Log effort Survey year Regenerating shrubs Brown antechinus 0.00 0.00 0.79 0.43
Landscape context Site Log effort Survey year Area Eastern chestnut mouse 0.00 0.00 2.88 0.00 Site Log effort Survey year Area Bush rat 0.00 0.00 -1.84 0.07 Site Log effort Survey year Area Swamp rat 0.00 0.00 -1.48 0.14 Site Log effort Survey year Area Brown antechinus 0.00 0.00 -2.21 0.03 Site Log effort Survey year Landscape heterogeneity Eastern chestnut mouse -0.36 0.18 -2.00 0.05 Site Log effort Survey year Landscape heterogeneity Bush rat 0.07 0.13 0.54 0.59 Site Log effort Survey year Landscape heterogeneity Swamp rat -0.99 0.35 -2.82 0.01 Site Log effort Survey year Landscape heterogeneity Brown antechinus 0.10 0.13 0.74 0.46
Pereoglou, F., Population processes in an early successional heathland 192
Appendix 5: Supporting information for Chapter 4
Supporting information for Chapter 4 Landscape genetics
Pereoglou, F., Population processes in an early successional heathland 193 Appendix 5:Supporting information
Appendix 5: Supporting Information for Chapter 4
Landscape genetics of an early successional specialist in a disturbance-prone environment F. Pereoglou, S. C. Banks, C. MacGregor, F. Ford, J. Wood and D. B. Lindenmayer
Appendix S1 (Methods for Structure and Geneland analyses)
In STRUCTURE, we ran five iterations of eight possible models (K=1-8) to obtain an average probability value for each model. We used the default parameter settings (admixture model, allele frequencies correlated) with a 200000 burn-in period and 1000000 MCMC repetitions after burn-in.
In GENELAND, we performed five independent runs with 100000 MCMC iterations (thinning=100) allowing K to vary from 1 to 20. We used a spatial model with 100 m error on coordinates and correlated allele frequencies. We inferred K from the modal value of the run with the highest likelihood.
Appendix S2 (Results for Structure and Geneland analyses)
The estimated logarithm of likelihood for data analysed in STRUCTURE was highest for K = 3 in both years (Fig. S1). When individuals were assigned to population clusters using probability values > 0.90, only 22 (2008-09) and 19 (2009-10) individuals could be assigned to a cluster, with the remaining 87 % of animals not able to be assigned. When a more stringent reclassification was applied using probability values > 0.95, no animals could be assigned to a population cluster in either survey year. This suggests that the three clusters identified in each year were not demographically discrete groups.
Using a spatial model in GENELAND, nine clusters were inferred in 2008-09 and eight in 2009-10 (Table S1). Individuals were not uniformly distributed to each of the identified population clusters (Table S2) and there was no spatial separation between populations, with individuals trapped at one site belonging to one or more populations. None of the probability levels upon which the assignments were made were greater than 90 % suggesting that the spatial model was also unable to detect demographically discrete groups corresponding to spatially segregated regions.
Pereoglou, F., Population processes in an early successional heathland 194 Appendix 5:Supporting information
Table S1: Number of sites surveyed, sites with eastern chestnut mouse captures, and occupancy by vegetation type. Unpublished annual monitoring data (see Lindenmayer et al. 2008)
Vegetation type Number of sites Number of sites Occupancy (%) surveyed with captures Forest 24 0 0 Rainforest 8 0 0 Woodland 23 4 17.4 Heathland (dry) 19 11 57.9 Heathland (wet) 7 4 57.1 Shrubland 20 1 5 Sedgeland 8 3 37.5 Total 109 23
Table S2: PCR conditions and characteristics of microsatellites. Number of alleles, expected heterozygosity estimated across sites combined over years
Locus Denaturation Annealing Extension # cycles # alleles HE temperature temperature (s) 2G6 95 61 – 51 60 35 4 0.7399 1C1 95 61 – 51 60 35 5 0.5825 6D4 95 61 – 51 60 40 3 0.3071 10E12 95 61 – 51 60 40 3 0.4368 5B11 95 61 – 51 60 40 6 0.6162 9A8 95 61 – 51 60 40 3 0.3998 7D12 95 61 – 51 60 40 3 0.5479 6E8 94 66 – 60 45 35 5 0.6697 1A7 94 66 – 60 45 35 4 0.6626 10G6 94 66 – 60 45 35 4 0.6523
Table S3: Pairwise site FST for the eastern chestnut mouse (FST for 2008-09 below the diagonal, 2009-10 above). All loci are combined. na indicates sites that were excluded from analysis. Non-significant values highlighted in bold
7-2 7-44 7-44B 7-46 7-95 7-H 7-I 7-M 7-P 7-2 na 0.106 0.062 0.016 0.269 0.273 0.183 0.027 7-44 0.244 na na na na na na na 7-44B 0.190 0.119 0.136 0.126 0.246 0.264 0.185 0.089 7-46 0.095 0.094 0.167 0.058 0.249 0.248 0.187 0.006 7-95 Na na na na 0.279 0.376 0.170 0.071 7-H 0.298 0.200 0.256 0.223 0.360 0.162 0.263 7-I 0.205 0.049 0.097 0.119 0.229 0.268 0.200 7-M Na na na na na na na 0.138 7-P 0.065 0.099 0.101 0.006 na 0.192 0.087 na
Pereoglou, F., Population processes in an early successional heathland 195 Appendix 5:Supporting information
Table S4: Results of clustering analysis in GENELAND. The value of K estimated for each of the five runs is presented with log-scale likelihood in decreasing ranked order.
20082008--09 20092009--1010 LikelihoodLikelihood K LikelihoodLikelihood KK -4684.793342 9 -4201.876479 88 -4718.412455718.412455 9 -4248.043092 88 -4929.102147 8 -4299.593687 88 -5011.304671 9 -4330.276253 88 -5163.609895 9 -4492.967675 88
Table S5: The number of individuals assigned to each population cluster in GENELAND.
2008-09 2009--1010 cluster nn clustercluster nn 1 33 11 4444 2 1212 22 3636 3 11 33 55 4 1616 44 4040 5 3333 55 11 6 2121 66 77 7 66 77 55 8 5757 88 33 9 2323
Pereoglou, F., Population processes in an early successional heathland 196 Appendix 5:Supporting information
Table S6: Results of sex-bias spatial autocorrelation analysis for both survey years. Male and female genetic correlation (r), upper (Ur) and lower (Lr) bootstrap 95% confidence limit for r, p-value for test r > 0 (Pr), T2 and probability (P) of sex-bias at spatial scales of up to 250 m and 5 km. Distance class in metres
Female Male Dist. class r Lr Ur Pr r Lr Ur Pr T2 P 2008-09 Fine-scale correlograms 0-50 0.246 0.196 0.297 0.001 0.162 0.103 0.223 0.001 4.182 0.033 50-100 0.145 0.103 0.188 0.001 0.150 0.109 0.189 0.001 0.021 0.894 100-150 0.117 0.067 0.172 0.001 0.167 0.116 0.219 0.001 1.707 0.202 150-200 0.053 -0.009 0.109 0.051 0.001 -0.067 0.073 0.522 1.114 0.284 200-250 0.088 0.006 0.167 0.025 0.125 -0.010 0.242 0.053 0.180 0.679 Coarse-scale correlograms 0-500 0.130 0.110 0.153 0.001 0.130 0.106 0.154 0.001 0.001 0.977 500-1000 0.063 0.028 0.097 0.001 0.077 0.041 0.113 0.001 0.253 0.620 1000-1500 0.002 -0.035 0.040 0.453 -0.001 -0.029 0.029 0.557 0.009 0.932 1500-2000 0.028 0.005 0.051 0.008 0.013 -0.010 0.037 0.109 0.731 0.399 2000-2500 -0.003 -0.027 0.022 0.642 -0.005 -0.033 0.024 0.682 0.011 0.914 2500-3000 0.010 -0.014 0.035 0.193 -0.025 -0.058 0.007 0.941 2.747 0.101 3000-3500 -0.061 -0.079 -0.043 1.000 -0.050 -0.069 -0.031 1.000 0.674 0.407 3500-4000 -0.028 -0.054 0.000 0.967 -0.076 -0.106 -0.048 1.000 5.383 0.018 4000-4500 -0.042 -0.095 0.010 0.961 -0.043 -0.087 -0.001 0.960 0.001 0.987 4500-5000 -0.132 -0.160 -0.104 1.000 -0.023 -0.051 0.003 0.964 29.418 0.001
2009-10 Fine-scale correlograms 0-50 0.191 0.126 0.256 0.001 0.207 0.125 0.294 0.001 0.087 0.751 50-100 0.202 0.158 0.247 0.001 0.063 0.009 0.118 0.008 12.531 0.001 100-150 0.206 0.149 0.262 0.001 0.142 0.078 0.213 0.001 2.060 0.141 150-200 0.088 0.023 0.163 0.036 -0.011 -0.094 0.092 0.566 2.158 0.142 200-250 -0.377 0.000 0.000 0.935 -0.036 -0.125 0.040 0.646 5.219 0.035 Coarse-scale correlograms 0-500 0.182 0.154 0.210 0.001 0.097 0.064 0.131 0.001 14.258 0.001 500-1000 0.082 0.034 0.129 0.001 0.037 -0.012 0.092 0.080 1.520 0.225 1000-1500 0.025 -0.009 0.059 0.085 0.027 -0.006 0.063 0.077 0.007 0.940 1500-2000 0.091 0.056 0.127 0.001 0.016 -0.020 0.050 0.175 7.363 0.008 2000-2500 0.004 -0.027 0.033 0.370 0.026 0.000 0.054 0.015 1.098 0.319 2500-3000 -0.007 -0.043 0.031 0.629 -0.006 -0.038 0.029 0.651 0.001 0.983 3000-3500 -0.061 -0.086 -0.041 1.000 -0.011 -0.035 0.012 0.855 7.860 0.006 3500-4000 -0.065 -0.097 -0.031 1.000 -0.014 -0.049 0.022 0.789 3.916 0.045 4000-4500 -0.071 -0.104 -0.040 1.000 -0.060 -0.093 -0.026 1.000 0.205 0.648 4500-5000 -0.058 -0.084 -0.032 1.000 -0.065 -0.096 -0.032 1.000 0.118 0.762
Pereoglou, F., Population processes in an early successional heathland 197 Appendix 5:Supporting information
Table S5: Results of sex-biased dispersal test for both survey years using FSTAT. Dispersing sex expected to have lower mean of assignment index and FST, greater variance of assignment index, FIS and HS compared to the philopatric sex
Male Female p Survey-09 year Test Male-0.079 Female0.068 p0.318 2008-09 Mean corrected assignment index 4.149-0.079 3.3740.068 0.1550.318 Variance of corrected assignment index 0.0974.149 0.0933.374 0.5250.155 FST -0.0250.097 -0.0690.093 0.1640.525 FIS 0.540-0.025 0.545-0.069 0.6300.164 -10 HS -0.1980.540 0.1700.545 0.1540.630 2009-10 Mean corrected assignment index 4.405-0.198 3.8910.170 0.3040.154 Variance of corrected assignment index 0.0864.405 0.1283.891 0.1540.304 FST -0.0140.086 -0.0060.128 0.4830.154 FIS 0.548-0.014 0.516-0.006 0.0740.483 HS 0.548 0.516 0.074
Figure S1. Number of population clusters identified using STRUCTURE. Data is mean posterior probability (LnP(D)) ± standard error based on 1000000 MCMC repetitions. Solid symbol is 2008-09 survey year (K = 3); hollow symbol is 2009-10 survey year (K = 3).
Pereoglou, F., Population processes in an early successional heathland 198 Appendix 5:Supporting information
Figure S2. The 95% highest posterior density intervals of GESTE population-specific FST estimates (x- axis) for the nine populations with the largest sample sizes (y-axis).
Pereoglou, F., Population processes in an early successional heathland 199
THE END