1 General Introduction

Home , African fish eagle, Australian magpie

A Multi-Scale Approach to Defining Historical and Contemporary Factors Responsible for the Current Distribution of the White-bellied Sea-Eagle Haliaeetus leucogaster (Gmelin, 1788) in Australia

Author Shephard, Jill

Published 2004

Thesis Type Thesis (PhD Doctorate)

School Australian School of Environmental Studies

DOI https://doi.org/10.25904/1912/3192

Downloaded from http://hdl.handle.net/10072/367440

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

Appendices 2 and 3 of this thesis consist of reproductions of previously-published journal articles, and have been removed from the electronic version of the thesis.

Citations for these articles can be found on page vii.

A MULTI-SCALE APPROACH TO DEFINING HISTORICAL AND

CONTEMPORARY FACTORS RESPONSIBLE FOR THE CURRENT

DISTRIBUTION OF THE WHITE-BELLIED SEA-EAGLE HALIAEETUS

LEUCOGASTER (GMELIN, 1788) IN AUSTRALIA

Jill Shephard B.Ed., B.Sc.(Hons)

Australian School of Environmental Studies

Faculty of Environmental Sciences

Griffith University

Submitted in fulfillment of the requirements of the degree of Doctor of Philosophy

September 2003

SYNOPSIS

The White-bellied Sea-Eagle Haliaeetus leucogaster is widespread in Australia, but has been the subject of conservation concern due to suggested localised declines and extinctions. Regionalised monitoring programmes have addressed some aspects of local concern, however a broader approach is needed to gain an understanding of large-scale processes affecting long-term persistence at scales equivalent to the species Australian range.

Ultimately, the ability to predict change in population size over time accurately depends on the scale of analysis. By necessity, ecological studies using direct sampling techniques are often made across spatial scales smaller than a species geographic range and across relatively short time frames. This seems counter-intuitive considering that long-term species persistence is often dependent on large-scale processes.

The principal aim of this thesis was to identify historical and contemporary forces responsible for the current pattern of population structure in H. leucogaster. This required a multi-scale approach, and the resulting research uses genetic, distributional and morphometric data.

Haliaeetus leucogaster is a large territorial raptor that historically has been associated with coastal regions, lakes and perennial river systems. It has an extensive worldwide

distribution from the western coast of India throughout the Indomalaysian region,

Papua New Guinea and Australia. By virtue of the species’ large-scale distribution, in

Australia it is fairly cosmopolitan in its use of habitat and prey types.

Haliaeetus leucogaster is monomorphic for adult plumage colouration, but in body size displays reversed sexual dimorphism with female birds significantly larger. A discriminant function based on 10 morphometric characters was 100% effective in discriminating between 19 males and 18 females that had been sexed using molecular genetic methods. Re-classification using a jackknife procedure correctly identified

92% of individuals. The discriminant function should be a viable alternative to genetic sexing or laparoscopy for a large proportion of individuals within the

Australo-Papuan range of this species; and can also be used to identify a small proportion of “ambiguous” individuals for which reliable sexing will require those other techniques.

I used mitochondrial (mtDNA) control region sequence data to investigate the current distribution of genetic variation in this species at the continental level and within and between specified regional units. I was specifically interested in identifying breaks in genetic connectivity between the west and east of the continent and between

Tasmania and the Australian mainland. Overall, genetic diversity was low and there was no significant level of genetic subdivision between regions. The observed genetic distribution suggests that the population expanded from a bottleneck approximately

160 000 years ago during the late Pleistocene, and spread throughout the continent through a contiguous range expansion. There is insufficient evidence to suggest

division of the population into different units for conservation management purposes based on the theoretical definition of the ‘evolutionary significant unit’. It is clear from the analysis that there are signatures of both historical and contemporary processes affecting the current distribution. Given the suggestion that population expansion has been relatively recent, additional sampling and confirmation of the perceived pattern of population structure using a nuclear marker is recommended to validate conservation monitoring and management at a continental scale.

To determine the existence of perceived population declines across ecological time scales, I analysed the Australian Bird Atlas Data to identify the extent and pattern of change in range and density of the species between three Atlas Periods (1901-1976,

1977-1981 and 1998-2001) using a new standardised frequency measure, the

Occupancy Index (OI) for 1° blocks (approx. 100km2) across the continent. At the continental scale, there was no significant difference in the spatial extent of occupancy between Atlas Periods. However, there were considerable changes in frequency and range extent between defined regions, and there were distinct differences in the pattern of change in OI between coastal and inland blocks over time. Coastal blocks showed much more change than inland blocks, with a clear increase in the use of coastal blocks, accompanied by a decrease in inland blocks, during the 1977 – 1981 Atlas Period, relative to both other Atlas Periods. The over- riding factor associated with distributional shifts and frequency changes was apparently climatic fluctuation (the 1977 – 1981 period showing the influence of El

Niño associated drought). The impression of abundance was strongly dependent on both the temporal and spatial scale of analysis.

iii

To test for correspondence between geographic variation in morphology and geographic variation in mtDNA I analysed morphometric data from 95 individuals from Australia and Papua New Guinea. First, the degree of morphometric variation between specified regions was determined. This was then compared with the pattern of genetic differentiation. There was a strong latitudinal cline in body dimensions.

However, there was no relationship between morphometric variation and patterns of genetic variation at least for mtDNA. Females showed a pattern of isolation by distance based on morphometric characters whereas males did not. Three hypotheses to explain the pattern of morphometric variation were considered: phenotypic plasticity, natural selection and secondary contact between previously isolated populations. I conclude that the pattern of morphometric variation is best explained by the suggestion that there is sufficient local recruitment for natural selection to maintain the observed pattern of morphometric variation. This implies that gene flow may not be as widespread as the mtDNA analysis suggested. In this instance either the relatively recent colonisation history of the species or the inability of the mtDNA marker to detect high mutation rates among traits responsible for maintaining morphometric variation may be overestimating the levels of mixing among regions.

As might be expected given the physical scale over which this study was conducted, the pattern of genetic, morphometric and physical distribution varied dependent on the scale of analysis. Regional patterns of genetic variation, trends in occupancy and density and morphometric variation did not reflect continental patterns, reinforcing

the contention that extrapolation of data from local or regional levels is often inappropriate.

The combined indirect methodologies applied in this study circumvent the restrictions imposed by direct ecological sampling, because they allow survey across large geographic and temporal scales effectively covering the entire Australian range of H. leucogaster. They also allow exploration of the evolutionary factors underpinning the species’ current distribution.

ACKNOWLEDGEMENTS

Foremost thankyou to my supervisors Jane Hughes, Carla Catterall and Penny Olsen for support, insightful criticism of drafts of this thesis and advice along the way.

For both assistance and logistical support in the field thankyou to Mia Hillyer, Jason Wiersma, Wen and Julie Nermut, Anne Williams, Janelle Ende, members of DNRE (Bairnsdale) and CALM (Kalbarri and Broome), and Nick Mooney (TPWS).

Many people contributed feather samples from private collections and I am deeply indebted to them for their generosity. Similarly, to Mike Double for sharing his feather extraction protocol and Glenn Graham for lab advice in the early days. Blood samples and morphometric data from Singapore were provided by the Jurong Bird Park (Singapore) and The National Bird of Prey Centre (United Kingdom). Access to skins and/or morphometric information was provided by curators at the Queensland Museum, Victorian Museum, Tasmanian Museum, South Australian Museum, Northern Territory Museum, Western Australian Museum, The National Wildlife Collection (CSIRO – Canberra), Currumbin Bird Sanctuary, Greenough Wildlife Park, Territory Wildlife Park, DNRE (Bairnsdale), CALM (Geraldton), Peter Frater, Nick Mooney (TPWS) and Jason Wiersma.

Thanks also to Birds Australia for providing the bird atlas base data, to Fiona Redfern (EPA – Qld) and Rose Eckleton (ESRI) for useful discussions over ARCVIEW 3.2, and to Michael Arthur and Rachel King for statistical advice.

Various members of ‘the lab’ provided advice with protocols, discussion over results or useful feedback on draft papers arising from this study. In particular I would like to thank Jing Ma, David Hurwood, David Gopurenko, Mia Hillyer, Snezana Dukic and Sonja Parsonage. Thankyou also to the Science faculty for running my sequences and to Petney Dickson and Roberta Bryant for administrative support.

I was lucky enough to be funded for the majority of my candidature through a Griffith University Postgraduate Research Scholarship and a Completion Assistance Postgraduate Scholarship. Field and laboratory work was supported in part through grant monies from The Norman Wettenhall Foundation, The Australian Bird Environment Foundation, The Australian Geographic Society, The Joyce W. Vickery Scientific Research Award, and The Stuart Leslie Bird Research Award.

To my family and friends, thankyou for your constant support and encouragement.

To Mia…….what could I possibly say.

For my Mum………………..carpe diem

PUBLICATIONS THAT HAVE ARISEN FROM THIS WORK.

Published

Wiersma JM, Nermut W, Shephard JM * (2001) A variation on the 'noosed fish' method and its suitability for trapping the White-bellied Sea-Eagle Haliaeetus leucogaster. Corella, 25 (4), 97-99.

(Trap refinement – all authors Paper preparation – JM Shephard (100%) * corresponding author)

Shephard, J.M., Catterall, C.P. & Hughes, J.M. 2004 Discrimination of sex in the White- bellied Sea-Eagle Haliaeetus leucogaster using genetic and morphometric techniques. Emu: Austral Ornithology 104 (1), 83-87.

In press

Shephard JM, Catterall CP, Hughes JM Long-term variation in the distribution of the White-bellied Sea-Eagle (Haliaeetus leucogaster) across Australia. Austral Ecology.

Accepted for Publication with review

Shephard JM, Hughes JM, Catterall CP, Olsen PD Conservation status of the White- bellied Sea-Eagle Haliaeetus leucogaster in Australia using mtDNA control region sequence data. Conservation Genetics.

Shephard JM, Hughes JM, Catterall CP Evidence of a morphometric cline in the White-bellied Sea-Eagle in the absence of mitochondrial DNA structure. Condor.

vii

TABLE OF CONTENTS

Synopsis i

Acknowledgements vi

Publications that have arisen from this work. vii table of contents viii list of Figures xi list of Tables xiii

Declaration xv

Chapter 1 - General Introduction 1

1.1 The impact of scale on analyses of population dynamics ...... 1 1.2 Application of macroecological techniques to answering questions in avian ecology ...... 4 1.2.1 The role of mitochondrial DNA in studying population structure...... 4 1.2.2 The analysis of abundance/occupancy relationships over ecological timescales...... 8 1.3 The White-bellied Sea-Eagle: ecology and conservation ...... 11 1.4 Overall aims of the study ...... 15

Chapter 2 – The distribution of genetic variation in the White-bellied Sea-Eagle in Australia 20

2.1 Introduction...... 20 2.2 Sampling Methods ...... 24 2.2.1 Sampling regime and choice of sampling regions ...... 24 2.2.2 Genetic Methods ...... 26 2.2.2.1 Tissue Treatment and Storage...... 26 2.2.2.2 DNA Extraction methodology ...... 30 2.2.2.3 Choice of gene region ...... 33 2.2.2.4 Description of isolation of control region and primer design ...... 33 2.2.2.5 PCR conditions ...... 36 2.2.3 Data analysis ...... 37 2.3 Results...... 41

viii

2.3.1 Variation within the fragment ...... 41 2.3.2 Evidence for genetic structuring within Australia ...... 44 2.3.3 Evidence for a historical bottleneck and population expansion...... 51 2.4 Discussion ...... 55 2.4.1 Reconstruction of colonisation history ...... 60

Chapter 3 – Variation in the distribution of the White-bellied Sea-Eagle across Australia at ecological time-scales 63

3.1 Introduction...... 63 3.2 Methods...... 64 3.2.1 Source of the Data...... 64 3.2.2 Treatment of the Data ...... 66 3.2.3 Data Analyses ...... 70 3.3 Results...... 73 3.4 Discussion ...... 82 3.4.1 Detecting ‘macro-scale’ changes in frequency and distribution...... 82 3.4.2 The effect of climatic variability...... 84 3.4.3 Human impacts and long-term trends ...... 86

Chapter 4 - Discrimination of sex using genetic and morphometric techniques 89

4.1 Introduction...... 89 4.2 Methods...... 90 4.2.1 Data collection ...... 90 4.2.2 Genetic sexing...... 92 4.2.3 Data analysis ...... 93 4.2.4 Results...... 95 4.3 Discussion ...... 101

Chapter 5 – Evidence of a morphometric cline in the White-bellied Sea-Eagle in the absence of mitochondrial DNA structure 105

5.1 Introduction...... 105 5.2 Methods...... 106 5.2.1 Morphometric analysis...... 106 5.2.2 Test of genetic and morphometric relationship...... 112

5.3 Results...... 115 5.4 Discussion ...... 122

Chapter 6 – General Discussion 128

6.1 Monitoring population fluctuation: the issue of scale...... 129 6.2 Evidence of population decline from the data?...... 131 6.3 Evidence for an alternate pattern of population structure ...... 134 6.4 Expansion of genetic analysis outside the Australian range of H. leucogaster...... 137

References 139

Appendices 164 Appendix 1 Source of Museum Specimens……………………………………..166

Appendix 2 Published Paper A variation on the 'noosed-fish' method and its suitability for trapping the White-bellied Sea-Eagle Haliaeetus leucogaster.

Appendix 3 Published Paper Discrimination of sex in the White-bellied Sea-Eagle Haliaeetus leucogaster using genetic and morphometric techniques.

LIST OF FIGURES

Figure 1.1 Total range distribution of the H. leucogaster...... 11

Figure 1.2 Typical undisturbed habitat used by H. leucogaster...... 13

Figure 2.1 The complete noosed fish design...... 25

Figure 2.2 a) Location of sample regions (shaded areas) superimposed across state boundaries; b) major fluvial divisions in Australia and the Carpentaria and Nullarbor Plain barriers ...... 27

Figure 2.3 a) The standard avian gene order compared to b) the novel gene order found in Haliaeetus leucogaster ...... 34

Figure 2.4 Frequencies of individual haplotypes from each region assayed...... 43

Figure 2.5 Cladogram showing the evolutionary relationship between haplotypes. Pies indicate the relative proportion of occurrence of haplotypes between regions with pie size showing the relative abundance of each haplotype..... 47

Figure 2.6 Scattergram for coastal and straightline distance against Slatkin’s linearised ΦST using Ln + 1 transformed values...... 50

Figure 2.7 The observed and expected distribution of pairwise differences used to test for deviation from the sudden expansion model...... 51

Figure 2.8 Nested clade diagram derived from the original cladogram...... 53

Figure 2.9 Distribution of 1-step clades within Australia. Of particular note is the cosmopolitan distribution of clade 1-1, the only clade other than the total cladogram to show a significant association between haplotype and location...... 55

Figure 3.1 a) Distribution of H. leucogaster sightings across all Atlas Periods within the Australian Bird Atlas Data; b) Regions used in the analysis and grid blocks in which there were ≥ 30 sheets for all three periods...... 68

Figure 3.2 Location of ‘urbanised’ (N=13) relative to ‘non-urbanised’ (N=110) coastal blocks...... 71

Figure 3.3 Frequency distribution of Occupancy Index values (%) per block across all Atlas Periods...... 72

Figure 3.4 Occupancy Index per 1° block for each Atlas Period. Pie charts quantify the proportion of blocks within each category...... 74

Figure 3.5 Occupancy Index (mean ± s.e.) within regions for each Atlas Period, and for coastal and inland locations...... 76

Figure 3.6 Distribution of the change in Occupancy Index within each Time Period comparison. Pie charts quantify the proportion of blocks to have increased, decreased or remained stable over time, and at each spatial scale...... 79

Figure 3.7 The change in Occupancy Index (mean ± s.e.) between Time Periods, within regions and for coastal and inland locations...... 80

Figure 3.8 Blocks colonised between the Historical Atlas (1901-1976) and Atlas 2 (1998-2001) Periods, and blocks in which H. leucogaster have apparently become locally extinct (i.e. an OI > 0% in the Historical Atlas but zero in Atlas 2). Blocks are shown relative to dams and other impoundments...... 81

Figure 4.1 Digested PCR product showing banding patterns for female (F: N=7) and male (M: N=5) H. leucogaster. Single lower bands (x: N=2) show failed PCR attempts relative to the male banding pattern. Sub-banding in the female is expected as it carries one z-band. The molecular weight standard (std) is pUC19...... 95

Figure 4.2 a) Plot of principal component scores for male and female birds using the first (Prin 1) and second principal component (Prin 2). b) Scaled vector loadings for each variable. Abbreviations are listed in Table 4.1...... 98

Figure 4.3 Plot of discriminant function scores against latitude for birds of known sex...... 100

Figure 4.4 Principal component scores (Prin 1: mean ± s.e.) for male and female birds for each latitude group...... 101

Figure 5.2 Male and female measurements (mean ± s.e.) for each morphometric variable in which there was significant variation between latitude groups...... 116

Figure 5.3 Distribution of Principal Component scores for female H. leucogaster relative to a) latitude groups, and b) regions...... 118

Figure 5.4 Distribution of Principal Component scores for male H. leucogaster relative to a) latitude groups, and b) regions...... 119

Figure 5.5 Principal component scores (mean ± s.e.) for a) female and b) male birds for each region...... 120

xii

LIST OF TABLES

Table 2.1 Summary of the numbers of different tissue types obtained from each region including the percentage of individuals for which multiple tissue types were available...... 29

Table 2.2 Variable sites for 399 bp of CR1 and 163 bp of CR II including two repeat units between position 179 and 198. The second repeat was removed from the analysis as it was not considered phylogenetically useful. Dots indicate identity to the concensus sequence HL1...... 42

Table 2.3 Mitochondrial DNA control region haplotype and nucleotide diversity within regions in Australia and outside Australia including neutrality tests.45

Table 2.4 Pairwise ΦST between each geographic region...... 46

Table 2.5 Frequencies of individual haplotypes from each region assayed...... 48

Table 2.6 Φ statistics and the hierarchical partitioning of genetic variation within and between pooled regions calculated from AMOVA using the distance method of Tamura & Nei (1993) and a gamma value of 0.09. Standard FST calculations are given in brackets. The theoretical movement of females between groups per generation is also included...... 49

Table 2.7 Nested clade analysis summary table showing the only clades with a significant permutational chi-square probability following 5000 re-samples...... 54

Table 3.1 The Occupancy Index compared using 2 factor ANOVA over time (a. Atlas Periods, 3 levels; b. Time Periods, 2 levels) and space (either Region, 8 levels or Location, Inland / Coastal). Tasmania was removed from the inland analyses due to insufficient replication...... 77

Table 3.2 The percentage of coastal blocks (N=123) within each Atlas Period with high Occupancy Index values (≥18%) subdivided into three categories...... 81

Table 4.1 Description of morphometric variables measured...... 91

Table 4.2 Summary statistics used to standardise raw measurements, correlations between morphological variables and principal components (Prin 1, Prin 2), and loadings for each variable in the discriminant function analysis (DFA)...... 99

xiii

levels) using two factor ANOVA. One-way ANOVAs were performed on Female and Male birds separately where the sex x latitude group interaction was significant (P <0.05) to determine if mean values varied significantly between latitude groups. Mean values for male and female birds are given separately. Values in brackets are standard errors...... 108

Table 5.2 Morphometric and genetic distance between regions for male and female H. leucogaster. The upper matrix is morphometric distance calculated as euclidean distance. The lower matrix is genetic distance based on Slatkin’s Linearised ΦST...... 113

Table 5.3 Summary of the number of individuals from each region used to calculate the morphometric distance matrices. The number ‘shared’, describes those individuals common to the morphometric and genetic data sets...... 114

Table 5.4 Relationship between morphometric distance and either geographic or genetic distance using Mantel’s Test...... 121

Table 5.5 Loadings for the first and second Principal Components for each sex..... 121

xiv

DECLARATION

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

Jill Shephard

xv Chapter 1 General Introduction

Chapter 1 - General Introduction

1.1 The impact of scale on analyses of population dynamics

By necessity, ecological studies using direct sampling techniques are often made across spatial scales smaller than a species geographic range (Goodwin and Fahrig,

1998) and across relatively short time frames (Marzluff and Sallabanks, 1998a). This seems counter-intuitive considering that long-term species persistence is often dependent on large-scale processes. Whilst the fundamental unit of analysis is the individual (Goodwin and Fahrig, 1998), generalist species with broad resource capabilities have the potential to become widespread as well as locally abundant

(Gaston and Lawton, 1990), and their abundance will typically be greatest at the centre of their geographic range, with peripheral populations showing consistently lower density (Brown et al., 1995).

Furthermore, the probability of detecting a species is directly linked to the size and number of sampling units, and therefore the perception of species abundance is linked to the spatial scale of survey (Schoener, 1987; Gaston and Lawton, 1990; Franklin,

1999; Guo et al., 2000).

Trends in abundance may also be strongly related to fluctuating conditions within regions, that are uncorrelated between regions (Innes, 1998; Kirk and Hyslop, 1998), such that recognised declines at the local level may manifest as stability over broader geographic areas (Flather and Sauer, 1996). For example, climatic variation may

1 Chapter 1 General Introduction

impact on species at a range of spatial scales, but the long-term effect of fluctuating environmental conditions cannot be assessed without long-term data sets covering appropriate geographic scales. Therefore, excepting geographically restricted species, analysis at the continental scale, or encompassing the known range of the species, is very important as localised processes are not readily extrapolated across broader areas

(Flather and Sauer, 1996; Maurer and Villard, 1996; Turner et al., 2001).

Spatial and temporal scales are not independent, and the scale chosen in a particular study may strongly influence the eventual outcome of the research (Goodwin and

Fahrig, 1998). According to Levin (1986) the ability to predict population persistence is highest at the scale of the global population and lowest at the scale of the individual. The lowest unit of measurement appropriate for surveying change is the population. However, studies of population dynamics are often hampered by a lack of knowledge with respect to population structure, dispersal characteristics, size and behaviour.

Bird species pose particular challenges as their implied dispersal potential is often at odds with actual population structure due for example to specific behaviours such as nest-site philopatry (e.g. Forero et al., 1999). Whilst geographic structure in bird populations has been studied using a variety of methods including morphology (e.g.

Fitzpatrick and Dunk, 1999), song (e.g. Tubaro and Segura, 1995), plumage (Hughes et al., 2001) and molecular markers (Zink, 1997), different methods do not necessarily reveal the same patterns of structure. For example, in North America the redpoll finch is genetically homogenous but shows extensive plumage and size

2 Chapter 1 General Introduction

variation geographically (Seutin et al., 1995). Predictably, without the ability to differentiate populations correctly, the assessment of trends in abundance is difficult.

This argument can be circular (i.e. how do you define the correct scale of analysis if you do not understand the physical extent of populations). However, analysis at large ecological or macroecological scales (Brown and Maurer, 1989) can determine both the physical extent of populations and change in population dynamics over time. Here

I define the term ‘macroecology’ as the study of any aspect of a species biology or ecology across large geographic areas and/or periods of time. Although macroecological research has been criticised as being too general (Blackburn and

Gaston, 2002), it has a strong application within conservation planning by creating benchmark data against which future results may be compared, as well as defining those areas or populations requiring study at smaller scales.

Advances in population genetics research and the availability of distribution data from Bird Atlas surveys have provided the opportunity to investigate aspects of a species biology at a range of spatial and temporal scales. Specifically, the use of genetic analyses in conjunction with time series analyses allows species patterns to be seen across a range of spatial and temporal scales. For example, an analysis of gene flow between regions can be used to define population extent, to identify important historical and contemporary processes affecting the current structure, and to survey genetic differences among populations over evolutionary time scales. On the other hand, analysis of trends in abundance using atlas data can be used to review shifts in abundance and occupancy over ecological time scales involving little or no change in

3 Chapter 1 General Introduction

gene frequency, and therefore movements or dispersal patterns that have occurred too recently to be seen within genetic data.

1.2 Application of macroecological techniques to answering questions

in avian ecology

1.2.1 The role of mitochondrial DNA in studying population structure

Storfer, 1996). However, it could be argued that a species’ ability to survive and reproduce hinges on its ability to disperse. Dispersal is inextricably linked to the movement of genes. Inferring levels of gene flow between populations gives an insight to the relative influences of selection and genetic drift on population structure

(Slatkin, 1987). Therefore an understanding of the historical and contemporary aspects of species dispersal and movement allows an understanding of the spatial patterning and interactions between wild populations.

Mitochondrial DNA (mtDNA) has emerged as a powerful tool in the study of gene flow among animal populations (Avise, 1994). The mitochondrial genome functions as a cytoplasmically inherited organellar genome, physically and genetically independent from the nuclear genome, but remaining functionally integrated with it

4 Chapter 1 General Introduction

(Brown, 1983). It is small and simple in comparison to the nuclear genome, is considered relatively simple to assay, displays a rapid rate of evolution, and is matrilinearly inherited (Brown, 1983; Harrison, 1989; Dowling et al., 1990; Avise,

1995). Notably, it exhibits considerable variation among individuals both within and between populations (Harrison, 1989), with variation generally being less within a population (Baverstock and Moritz, 1990). This property in particular has made mtDNA an effective marker of intraspecific variation and phylogeographic inference

(Moritz et al., 1987).

MtDNA has proven particularly versatile, as it can be used to answer questions of long-term significance such as genetic variability or the identification of evolutionary divergent populations, as well as questions of demographic importance to current conservation management such as the identification of hybrid zones (Harrison, 1989).

Additionally, studies of gene flow can be used to determine current population structure and therefore assist in the determination of appropriate scales for ecological monitoring or study. From a conservation perspective mtDNA analysis has a particularly important role. As a marker of female-mediated gene flow it can be used to infer female recruitment to an area. The ongoing persistence of a population is dependent on female recruitment. It is therefore an important tool in the estimation of recolonisation potential following short-term disturbance or local extinction (Rand,

1996; Taylor et al., 2000).

More recently, advances in molecular analysis have allowed the separation of historical and contemporary influences on genetic distribution. Traditional analysis of

5 Chapter 1 General Introduction

the geographic structure of genetic variation based on F-statistics (Wright, 1951), can only describe the current level of population structure (Templeton, 1998). Using an

FST approach, the proportion of variation among subpopulations is given a value between 0 and 1. An estimated FST value that is not significantly different from zero indicates panmixia, whereas an FST of one suggests complete divergence (Wright,

1951). Theoretically, a non-zero FST , which implies some degree of contemporary gene flow, requires the movement of only a single individual per generation (Avise,

1994). However, with recent colonisation events, an identical pattern of haplotype sharing can occur which may be the result of retention of ancestral haplotypes following the original expansion, rather than current gene flow (Zink, 1997). Using coalescent approaches (Crandall and Templeton, 1993), and nested clade analysis

(Templeton, 1998), it is possible to separate historical events such as population fragmentation or range expansion from contemporary patterns of gene flow, giving a more accurate insight to current dispersal ability or recolonisation potential.

Different parts of the mtDNA genome evolve at different rates. As such, specific sequences can be targeted to address questions of different temporal resolution. For example, the 12S and 16S ribosomal RNA genes, and part of the Cytochrome b gene, have a low mutation rate and evolve reasonably slowly, making them appropriate for distinguishing relatively distant species or genera (Erlich and Arnheim, 1992). In contrast, the control region is thought to evolve most rapidly, and is therefore most useful for, within rather than between species studies (Birley and Croft, 1986; Brown et al., 1986; Moritz et al., 1987). Its utility in determining species divergence is questionable as it is often subject to homoplasy. Under these conditions genetic

6 Chapter 1 General Introduction

similarity is acquired through independent evolutionary events and therefore does not necessarily reflect shared ancestry (Page and Holmes, 1998). Furthermore, this is more probable between distantly related taxa.

While early studies of mtDNA evolution estimated rates of around 2% divergence per million years for the complete genome (Brown et al., 1979), recent estimates of average divergence across Domain I, II and III of the avian mitochondrial control region are 20%, 5% and 23% per million years respectively (Baker and Marshall,

1997), making it particularly useful for defining closely related mitochondrial lineages within species. Rates of evolution may also be used to estimate the timing of evolutionary events, such as time since colonisation (Rogers and Harpending, 1992).

A number of authors have recommended caution with the use of mtDNA markers in the determination of priorities for species conservation (Avise, 1994; Moritz, 1994a) as mtDNA is effectively a single locus, subject to neutral evolution, and theoretically insensitive to the effects of selection (Avise et al., 1987; Rand, 1996). However, where the number of samples may be limited due to species rarity or logistic constraints, mtDNA can be very useful. As a haploid molecule with only maternal inheritance, a corresponding reduction in effective population size amplifies the effect of genetic drift (Birky et al., 1989). Therefore, if variation exists, differences will more readily be detected with mtDNA than with nuclear genes (Moritz, 1994a).

7 Chapter 1 General Introduction

1.2.2 The analysis of abundance/occupancy relationships over ecological timescales

The question of whether species are increasing or decreasing in abundance is of central importance in conservation research. Abundance indices (e.g. reporting rates), derived from census data, have been used in different forms in an increasing number of studies to quantify abundance, and change in abundance, over various spatial and temporal scales (e.g. Brown et al., 1995; James et al., 1996; Siriwardina et al., 1998;

Chamberlain et al., 2000; Guo et al., 2000), and are accepted as a useful diagnostic tool in identifying population trends in ecological time-series data. In this context a

‘time-series’ describes data taken at a single point in space in a series of independent sampling events thereby allowing comparative analysis. The most commonly cited census data in the bird literature come from the British Bird Survey (e.g. Chamberlain et al., 2000), the American Breeding Bird Survey (e.g. Kirk and Hyslop, 1998) and the Australian Bird Atlas Data (e.g. Garnett et al., 2002).

Atlas data usually involve direct counts or estimates of species abundance at a specified physical point in time and space. For example, the Australian Bird Atlas

Data comprises sheet records on which are listed species sightings within a specified sample time and location (Blakers et al., 1984).

There is a correlation between abundance and occupancy, where abundance is some measure of density and occupancy describes the percent occurrence in spatial units

(Gaston and Lawton, 1990). Therefore, large amounts of information can be inferred

8 Chapter 1 General Introduction

about population dynamic processes if atlas data can be quantified reliably. The power of such information is dependent upon the integrity of the original data

(Robertson et al., 1995), and there has been considerable discussion in the literature

(e.g. Thomas, 1996; Franklin, 1999) about which interpretive methods may be best for any given research problem. In most instances, different methodologies have been devised to counter the confounding influence of the many sources of error associated with census data (Flather and Sauer, 1996; Thomas, 1996; Kirk and Hyslop, 1998).

These sources of error include: survey bias toward individual species, differential observer effort between surveys and inadequate spatial coverage of a species range.

Time-series data can be used to identify and appropriately interpret anomalies within the overall trend (Jassby and Powell, 1990). Quantifying data at large geographic or temporal scales can have an overall smoothing effect, and depending on the scale of analysis may affect the quality of the final result. However, ‘smoothed data’ are useful in large temporal scale studies as they avoid attributing undue emphasis to short-term or fine-scale perturbations such as seasonal variation and environmental condition (Flather and Sauer, 1996), habitat variation (Thomas, 1996), or observer bias (Faanes and Bystrak, 1981; Link and Sauer, 1997) preserving only important increases or declines in abundance (Jassby and Powell, 1990; Barker and Sauer,

1992). For example Siriwardina et al. (1998) used smoothed index series trend data to identify key environmental disturbances in multi-species analysis. Without the use of the smoothing algorithm the underlying population trend was obscured by short-term fluctuations in abundance and measurement error.

9 Chapter 1 General Introduction

There are a number of disadvantages associated with atlas data, both across short and long time periods. For example, analysis over relatively short periods may produce a biased picture of abundance as the censused population is likely to be only a subsample of the overall population of interest (Thomas, 1996), populations may have been subject to short-term disturbance producing an underestimate of true abundance

(Innes, 1998), or inconsistencies in sample effort may under or over estimate abundance (Thomas, 1996). Over long temporal frames, it may be necessary to combine data from several atlas data sets. While this is potentially powerful, it can be particularly problematic as different atlases rarely adopt the same methodology. An interesting critique of this issue is provided by Blackburn & Gaston (2002).

According to Franklin (1999), data from separate atlases are combined on the assumption that records constitute an unbiased representation of the birds observed, that the geographic spread of records is the same and that observer effort is consistent.

Therefore, careful consideration must be given to the raw data to ensure that combined data or comparative analyses are valid.

This type of data may be incorporated within a geographic information system (GIS), and spatial and abundance information may be displayed concurrently. It is therefore possible to view distribution and occupancy between distinct time periods across large geographic extents, and statistically test change in both parameters, effectively allowing the description of trends over ecological time. Accordingly, GIS has become a powerful tool in the spatial analysis of wildlife distribution and abundance, particularly with respect to aiding the development of conservation strategies

(Aspinall, 1992; Brown et al. 1994; Ellis and Seal, 1995).

10 Chapter 1 General Introduction

1.3 The White-bellied Sea-Eagle: ecology and conservation

The White-bellied Sea-Eagle Haliaeetus leucogaster is a large territorial raptor that historically has been associated with coastal regions, lakes and perennial river systems (Favaloro, 1944; Marchant and Higgins, 1993). It has an extensive worldwide distribution from the western coast of India throughout the Indo-malaysian region, Papua New Guinea and Australia (Ferguson-Lees and Christie, 2001) (Fig

1.1).

Figure 1.1 Total range distribution of the H. leucogaster (Adapted from Ferguson– Lees and Christie, 2001)

11 Chapter 1 General Introduction

By virtue of the species’ large-scale distribution, in Australia it is fairly cosmopolitan in its use of habitat and prey types. Although generally associated with coastal areas and perennial river systems, it also breeds successfully on sea stacks, cliffs and escarpments on offshore islands where there are no suitable trees (Marchant and

Higgins, 1993; Ferguson-Lees and Christie, 2001) (Fig. 1.2).

Considered an opportunistic feeder, H. leucogaster hunts over deep water, coastal areas (including coral cays), terrestrial wetlands and lakes, and also inland terrestrial areas, to feed on all forms of aquatic vertebrates (including birds), reptiles, mammals, offal and carcasses (for a complete description of prey types see Marchant and

Higgins, 1993; Ferguson-Lees and Christie, 2001). Nest sites are usually selected in conjunction with their accessibility to water. This distance varies regionally and appears to be linked to environmental productivity and the availability of hunting perches (Thurstans, 1998), with distances ranging from less than 100m to several kilometers (pers. obs.; Emison and Bilney, 1982; Dennis and Lashmar, 1996;

Mooney, 1998; Thurstans, 1998; Williams, 1998a).

The home range over which H. leucogaster may forage is estimated to be large (up to

100km2) and proportional to the available food supply (Marchant and Higgins, 1993).

It is smaller during the breeding season. Birds are thought to pair for life, and retain a breeding territory for many years (Marchant and Higgins, 1993). Adults are considered sedentary within a territory although there are records of adult birds leaving the nest site after the young have fledged (Favaloro, 1944)

12 Chapter 1 General Introduction

North Queensland † Northern Australia †

Furneaux Group † (Bass Strait)

Cape Peron (Western Australia)**

Encounter Bay ‡ Tasman Peninsula † (South Australia)

Burrum River (Central Queensland)**

Hunter Island – Bass Strait*

Lakes Entrance (Victoria)**

Figure 1.2 Typical undisturbed habitat used by H. leucogaster; including some sampling sites from this study (Image sources: * N.Mooney; ** J.Shephard; ‡ J. Bracken; † Department of Environment and Heritage (2003)).

13 Chapter 1 General Introduction

There is limited information with respect to the dispersal of juveniles. They are capable of long distance movement, and there is an Australian Bird and Bat Banding

Scheme record of a juvenile having moved 3000km from its natal territory in South

Australia to Fraser Island in Queensland (Marchant and Higgins, 1993).

There is considerable regional variation in the spacing of nests. Estimates range from

1 pair/5-7km to 1 pair/65km (Olsen, 1995; Thurstans, 1998; Ferguson-Lees and

Christie, 2001) with more densely packed nesting territories (1 pair/1km) around man-made dams or other impoundments (pers. obs.). There are no estimates available for more remote locations within northern Australia. Whilst breeding pairs are often the only measure available to estimate population size (Olsen, 1995), raptors have a large ‘floating population’. That is, non-territorial, non-breeding adults of both sexes that live unobtrusively within existing territories (Olsen, 1995; Hunt, 1998). The number of territorial pairs may not be correlated with the number of reproducing birds, as breeding success, or potential, can vary from year to year depending on environmental and/or resource fluctuations (Olsen, 1995). Additionally, high juvenile mortality rates compound difficulties in estimating population sizes (Kenward et al.,

2000) and there are no reliable estimates available for the current population size of

H. leucogaster within Australia (Ferguson-Lees and Christie, 2001).

Like many other raptors, H. leucogaster has been adversely affected by human induced disturbance (Mooney and Brothers, 1986; Newton, 1991; Clunie, 1994;

Stokes, 1996). Although it is considered to have a declining world population, within

Australia the population is considered stable in the northern and remote tropical

14 Chapter 1 General Introduction

regions of its Australian range (Dennis and Lashmar, 1996). However, it is believed to have declined to the point of local extinction in parts of southern Australia (New

South Wales, Victoria and South Australia), and to be under considerable local pressure in Tasmania in response to various combinations of habitat clearance, human disturbance accidents and direct persecution (Emison and Bilney, 1982; Mooney and

Brothers, 1986; Marchant and Higgins, 1993; Dennis and Lashmar, 1996). It is listed as threatened in Victoria under the Flora and Fauna Guarantee Act 1988 (Olsen,

1995), and the total Victorian population is considered to be extremely restricted with only 100 breeding pairs estimated as remaining (Clunie, 1994). A number of studies have suggested that habitat modification and destruction pose a significant threat to

H. leucogaster through the removal of suitable nest trees (Emison and Bilney, 1982;

Bilney and Emison, 1983; Mooney and Brothers, 1986; Clunie, 1994; Dennis and

Lashmar, 1996). A study by Bilney and Emison (1983) within the Gippsland Lakes district of Victoria showed clearly that the number of successful nests and fledged young were many orders of magnitude less from nest sites within sub-optimal habitat, where sub-optimal habitat was defines as pasture containing isolated stands of trees

(Bilney and Emison, 1983).

1.4 Overall aims of the study

In order to address management and conservation priorities in any species it is essential to have an understanding of the fundamental aspects of its biology and ecology. For the White-bellied Sea-Eagle, regionalised monitoring programmes (e.g.

Williams, 1997, 1998a, b) have addressed some aspects of local concern. However, a

15 Chapter 1 General Introduction

broader approach is needed to gain an understanding of large-scale processes affecting long-term persistence of H. leucogaster at the scale of the species’

Australian range. The present study uses a multi-scale approach, together with genetic, distributional and morphometric data, to identify historical and contemporary factors responsible for the current pattern of population structure in this species.

Broadly, this study aims to: 1) determine the current genetic status of H. leucogaster within Australia, particularly with respect to levels of diversity and geographic structure, and 2) examine available atlas data to test the presence and/or extent of persistent decline in abundance or range extent within Australia. The nature of the data also made it necessary to develop a tool to allocate sex accurately in this species.

From a management perspective it was expected that new information from each component of the study would contribute to the determination of conservation need across a range of spatial scales within Australia, create a baseline against which future change in genetic, morphometric or physical distribution can be compared, and provide a statistically derived recommendation of an appropriate physical scale for ongoing conservation monitoring.

Chapter 2 considers the distribution of genetic variation in the White-bellied Sea-

Eagle in Australia using mitochondrial DNA. A species colonisation history may have a profound effect on current patterns of population structure. To date there is no information regarding intraspecific genetic variation in the White-bellied Sea-Eagle

(Marchant and Higgins, 1993; Ferguson-Lees and Christie, 2001). This chapter describes the current distribution of genetic variation in the species at the continental

16 Chapter 1 General Introduction

level and within and between specified regions with a view to collecting baseline genetic information that may assist with future conservation management and the determination of geographic population extent. Specifically the data are used to test whether the Australian distribution is comprised of distinct population units, or is effectively a single inter-breeding population. The data are also used to infer the effect of colonisation history on current genetic structure and the implications of this in a conservation management framework.

Chapter 3 assesses long-term variation in the distribution of the White-bellied Sea-

Eagle across Australia. Spatial analysis can be used to identify range movements or species contractions. In this chapter I make use of the Australian Bird Atlas Data to identify the extent and pattern of change in range or density of the White-bellied Sea-

Eagle in Australia over differing spatial and temporal scales. This work complements the genetic analysis as, although it is made across comparable geographic scales, it looks at variation across long-term ecological time-scales to determine whether ther are relatively short-term shifts in population structure that can not be seen using genetic analysis. This analysis also identifies important current processes affecting occupancy and abundance at the continental scale and between specified regions.

Chapter 4 develops a technique for discrimination of sex in the White-bellied Sea-

Eagle. A large number of questions in ecology and conservation depend on the ability to sex organisms correctly. Adult H. leucogaster are monomorphic for plumage colouration, but display reversed sexual size dimorphism, the female being moderately larger (Marchant, 1993). Where a male and female are seen together it is

17 Chapter 1 General Introduction

easy to assign sex, however where birds are observed individually it can be more difficult. To facilitate the use of appropriate sample sizes for the analysis of morphometric variation described in Chapter 5, it was necessary to develop a tool that could be used to reliably assign sex in this species. Accordingly, the aim of this chapter was to assess methods that can be used to determine sex both in live and museum specimens. First, I assessed the utility of genetic analysis using DNA from samples including museum specimens and shed feathers. The second aim was to assess the utility of disciminant function analysis to assign sex to birds of unknown sex on the basis of morphometric characters.

Chapter 5 assesses the geographical pattern of morphometric variation of the White- bellied Sea-Eagle in Australia. A large number of bird species show morphological variation along a latitudinal cline (Zink and Remsen, 1986). In many cases this variation does not agree with other measurements of population structure. This discrepancy has been attributed to a range of factors including: the recent evolution of morphometric variation, natural selection, secondary contact and environmental induction. Therefore, the aim of this chapter was to quantify morphometric variation in male and female birds using the same regional groups previously used to explore genetic and distributional change in the species. Specifically I was interested in determining if there was a relationship between morphometric and genetic variation and exploring some mechanisms that could be responsible for maintaining morphometric variation.

18 Chapter 1 General Introduction

Chapter 6 provides a synthesis of information from the previous chapters to recommend an appropriate geographic scale at which to monitor population change, and having adopted that scale, to determine if, based on the current data, there is statistical evidence of decline in H. leucogaster. Finally, some outstanding issues and future research directions are outlined.

19 Chapter 2 Genetic Distribution

Chapter 2 – The distribution of genetic variation in the White-

bellied Sea-Eagle in Australia

2.1 Introduction

Central to the determination of a species conservation status is an assessment of genetic variation within and between extant populations, as well as an investigation of the current and historical processes that have shaped, and may determine, the species’ long-term survivorship. Small population sizes are typically associated with threatened populations. The genetic consequences of a bottleneck due to population decline following a threatening process, or within a newly founded population following a colonisation event, are the same. In either case the extent to which genetic diversity is affected depends on the size of the ‘new’ population, the duration of the bottleneck and the effective number (Ne) of individuals contributing to ensuing generations (Chakraborty and Nei, 1977; Hodson, 1992; Stockwell et al., 1996).

Predictably, populations that have been subject to a bottleneck show evidence of a low historical Ne as indicated by low haplotype and nucleotide diversity (Glenn et al.,

1999; Mila et al., 2000). Additionally, the time since the bottleneck and rate of expansion following the bottleneck episode strongly determine genetic structure (Nei et al., 1975; Mila et al., 2000; Shephard et al., 2002).

Species range expansions and colonisation events occur as either an advancing wave of continuously distributed individuals, or through the establishment of isolated

20 Chapter 2 Genetic Distribution

populations via long-range dispersal with admixture at a later stage as newly created populations become established and more widespread (Nichols and Hewitt, 1994).

With recent colonisation events it is often difficult to distinguish between historical and contemporary signatures of gene flow, making it difficult to interpret the path of colonisation, or the current level of population structure (Zink, 1997). For example, individual haplotypes are often found in multiple populations. This could be evidence of contemporary gene flow, or the retention of ancestral haplotypes under non- equilibrium conditions (Zink, 1997). The inability to differentiate between the two can lead to inaccurate interpretations of genetic connectivity as well as the mechanisms determining the current level of structural partitioning (Slatkin and

Maddison, 1989; Edwards, 1993b; Avise, 1994; Mila et al., 2000).

Though moderated by species dispersal ability or behaviour, at large spatial scales, intraspecific studies on mammals have tended to find significant genetic variation between demes, with some degree of geographic structure (Ball et al., 1988). In contrast, genetic differentiation in avian species is documented as low (Ball et al.,

1988; Avise, 1996; Crochet, 2000; Mila et al., 2000). Even so, within avian groups, there are some classic examples of species showing virtually no genetic differentiation across broad geographic scales (e.g. the Red-winged Blackbird

Agelaius phoenceus; Ball et al., 1988), while others show clear phylogenetic division among mitochondrial DNA (mtDNA) lineages of spatially co-occurring individuals irrespective of low levels of overall diversity (Avise, 1996).

21 Chapter 2 Genetic Distribution

Australia is unique among the continents, bar the Antarctic, in that it is a discrete geographic entity completely surrounded by water. There is now significant information to suggest that at least some groups of passerine birds originated within

Australia, while other avian taxa are thought to have immigrated from either the north, or west through the Indo-Malaysian region (Kikkawa et al., 1981; Keast,

1981a, 1981b; Barker et al., 2002; Ericson et al., 2002). To date there have been few intraspecific molecular studies (see for example: Edwards, 1993a; Edwards, 1993b) assessing the effects of this evolutionary history on the genetic structure of bird populations at large ecological scales within Australia.

Based on phylogenetic reconstruction using mitochondrial DNA (cytochrome b and part of one tRNA gene), H. leucogaster forms part of a clearly defined monophyletic group within the Haliaeetus genus, and is most closely associated with the afro- tropical African Fish-Eagle H. vocifer (Seibold and Helbig, 1996). Haliaeetus leucogaster has an extensive distribution from the western coast of India throughout the Indo-Malaysian region, Papua New Guinea and Australia (Ferguson-Lees and

Christie, 2001). Its current distribution and phylogenetic position within the genus have been used to suggest that H. leucogaster most likely colonized Australia through

Southeast Asia (Seibold and Helbig, 1996).

Haliaeetus leucogaster is continuously distributed around coastal Australia and inland around permanent lakes and river systems (Blakers et al., 1984). There are no obvious current barriers to gene flow on mainland Australia (Ford, 1987; Olsen, 1995).

However, Tasmania is separated from the mainland by the Bass Strait, approximately

22 Chapter 2 Genetic Distribution

200km of open sea, which contains a few islands on which the species is known to have permanent breeding pairs. Only one coastal range break occurs, between western

Victoria and eastern South Australia. This break is relatively recent, potentially in response to habitat alteration as a consequence of agricultural expansion (AUSLIG,

2001). To date there is no information regarding the distribution of genetic variation at the intraspecific level (Marchant and Higgins, 1993; Ferguson-Lees and Christie,

2001).

I used mitochondrial control region sequence data to determine the current distribution of genetic variation in this species at the continental level, and within and between specified regional units. Specifically I aimed to: 1) determine the degree of genetic structure present within the species’ Australian range, 2) identify evidence of restricted gene flow between the Australian mainland and Tasmanian populations, or between the west and east of the continent, 3) collect baseline genetic information as a proactive approach to assist in identifying and/or setting future management priorities through the definition of distinct population units, 4) determine if the

Australian distribution was subject to a bottleneck at the time of colonisation.

Additionally, I propose hypotheses to explain the physical colonisation history of the species and explore some of the mechanisms that may have assisted its expansion to

Australia.

23 Chapter 2 Genetic Distribution

2.2 Sampling Methods

2.2.1 Sampling regime and choice of sampling regions

Birds of prey by their very nature are difficult to capture. Within Australia individuals were trapped using a noosed-fish technique. The noosed trap design has been used extensively in North America, and was first used to trap Bald Eagles in Alaska

(Robards, 1967). The technique used in this study is a modification of the noose system of Frenzel and Anthony described in Bloom (1987), and utilises a 4-noose system in lieu of the original 2-noose design (Fig 2.1). This design is suitable to both boat and land-based set-ups, and allows catching of adult and immature H. leucogaster away from the nest. This work was subsequently published and is reproduced in the Appendix.

Haliaeetus leucogaster has a history of sensitivity to nest site disturbance (Mooney and Brothers, 1986; Marchant and Higgins, 1993; Dennis and Lashmar, 1996), and whilst adults are more readily located within known territories, disturbance susceptibility precludes the trapping of adults or young at the nest for fear of unnaturally increasing chick mortality as a result of nest and/or territory desertion by adults. Live trapping of individuals took place in Tasmania, Victoria, South-East

Queensland, Central Queensland and Western Australia. To increase sample numbers, live caught samples were supplemented from museum collections, private collections including government agencies (e.g. CSIRO), and field collections from natural moult.

24 Chapter 2 Genetic Distribution

Figure 2.1 The complete noosed fish design. (a) Detail of the overhand knot used to form a self-regulating noose. (b) Diagram of the final water based set up.

25 Chapter 2 Genetic Distribution

Implicit within the aims of the study was the chance to determine if future genetic research could utilise museum and feather tissue, where live capture was not required.

This would have the added benefit of reducing the level of field based disturbance on natural populations of H. leucogaster.

Individuals were assigned to pre-defined geographic sample regions (Fig. 2.2a) corresponding loosely to the major fluvial divisions of Australia (Allen, 1989; Fig.

2.2b) or with accepted biogeographic boundaries (Thackway and Cresswell, 1995).

Additionally, for comparative purposes, a number of blood and museum samples were obtained from Singapore (N=3) and Papua New Guinea (N=4).

2.2.2 Genetic Methods

2.2.2.1 Tissue Treatment and Storage

Five different tissue types were used including: blood (N=25), feather (N=97), muscle

(heart and pectoral; N=3), ‘fresh’ tissue (liver and toe pad; N=2), and dried tissue (toe pad). Both blood and chest feathers were obtained from live caught birds. 100 µl of blood was taken from the brachial vein via sterile syringe and immediately added to1ml of lysis buffer (100mM Tris Hcl pH 8.0,100mM EDTA, 0.5% SDS, 10mM

NaCl), mixed thoroughly and stored at ambient temperatures until it could be transferred to a –70°C freezer for long term storage.

26 Chapter 2 Genetic Distribution

a) Singapore Region 7 43 20 km Papua New Guinea Region 8 340 0 km 720 km

2 Region 4 8 5 0 k m

Region 5 Region 3 N. T. QLD

W.A. S.A N.S.W .

Region 6 VIC. Region 2

Region 1 TAS.

b) CCaarpentrpentaaririaa bbreakreak

NNullarborullarbor PlainPlain

Figure 2.2 a) Location of sample regions (dotted circles) superimposed across state boundaries. Australian states are: WA – Western Australia, NT – Northern Territory, QLD – Queensland, SA – South Australia, NSW – New South Wales, Vic – Victoria, TAS – Tasmania. Points correspond to actual individual sample locations within regions. Regions are: 1 = Tasmania, 2= Southeast Australia, 3 = Central & Southeast Queensland, 4 = Northern Australia, 5 = Central Western Australia, 6 = South Australia, 7 = Singapore, 8 = Papua New Guinea; b) major fluvial divisions in Australia and the Carpentaria and Nullarbor Plains barriers purported to function as barriers to dispersal in some raptor species (Olsen, 1995). Map adapted from Allen (1989).

27 Chapter 2 Genetic Distribution

Muscle was obtained from various museum tissue banks and other government collections (e.g. CSIRO). In all cases the source material had been frozen at a minimum of –20°C, and was either stored as a dissection or dissected from a complete carcass. Sub-samples for the study were stored in lysis buffer and treated in the same manner as blood. Dissection was performed using a sterile scalpel for each specimen, with flame sterilisation of other equipment between specimens.

Feathers were also collected from museum skins, by direct plucking, and in the field following natural moult. These ‘dried’ feathers were transferred to individual plastic bags and later stored at –70°C. Similarly, dried skin was obtained from the toe pads of museum skins by dissecting approximately 3mm cubes with a sterile scalpel. This material was also stored in individual plastic bags and then at –70°C.

Where possible, multiple tissue samples (i.e. more than one type) were obtained from the same specimen to ensure identical results between tissue types, and to ensure the inter-changability of tissue types across the whole study, as the final analysis was performed based on the full range of tissues. Calibration of the source types was seen as an important issue as some studies have revealed the presence of mitochondrial insertions within the nuclear genome. These sequences may be preferentially amplified by some primer pairs leading to unreliable results (see for example:

Greenwood and Pääbo, 1999). A sub-sample of individuals (N=3) was screened to test for the preferential amplification of nuclear copies following the methods of Mila et al. (2000). That is, forward and reverse sequences were obtained for three individuals using a 1030bp segment of mitochondrial control region DNA. Sequences

28 Chapter 2 Genetic Distribution

were compared between blood and feather (one individual); blood, feather, liver, and dry toe pad (one individual); and feather and pectoral muscle (one individual).

Sequence alignments were concordant within individuals in all comparisons. It was assumed from this that primer pairs did not preferentially amplify nuclear copies, and all source tissue types were used in this study. A summary of source tissue types is presented in Table 2.1.

Table 2.1 Summary of the numbers of different tissue types obtained from each region including the percentage of individuals for which multiple tissue types were available. Regions correspond to those in Figure 4.1b where: 1 = Tasmania, 2= Southeast Australia, 3 = Central and Southeast Queensland, 4 = Northern Australia, 5 = Central Western Australia, 6 = South Australia, 7 = Singapore, 8 = Papua New Guinea.

Sample Type Regions

1 2 3 4 5 6 7 8 Total

Blood 11 2 4 4 3 1 3 28 Feather 45 41 31 11 19 17 8 172 “Fresh” tissue 2 3 1 1 7 Dried skin 32 8 13 7 17 5 7 80 Total samples of all types 90 54 49 22 39 24 3 15 296 Total No. of indiv per region 59 34 34 15 26 22 3 8 201 No. of indiv with > 1 tissue 30 10 11 8 5 2 0 7 73 type % of total with > 1 tissue type 50.9 29.4 32.4 53.3 19.2 9 0 87.5

No. of indiv sequenced 39 22 20 17 10 17 3 4 128

29 Chapter 2 Genetic Distribution

2.2.2.2 DNA Extraction methodology

Extraction protocols varied according to tissue type. Total DNA was extracted from blood following a standard phenol-chloroform protocol. Approximately 30µl of lysed blood, or a few scrapings of frozen blood, was added to 400µl of lysis buffer (100mM

Tris HCl pH 8.0,100mM EDTA, 0.5% SDS, 10mM NaCl) and 12.5µl of Proteinase

K (20mg/ml) in a 1.5ml Eppendorf tube, and incubated overnight on a rotating wheel at 37°C. The lysate was cleaned twice with an equal volume of phenol, including a 15 minute mixing phase with rotation. Once the solution was thoroughly mixed, it was centrifuged for 3 minutes and the aqueous phase pipetted to a new Eppendorf tube.

This procedure was repeated with 200µl of phenol and 200µl of chloroform-isoamyl alcohol (24:1) and then again with 400µl of chloroform-isoamyl alcohol (24:1). DNA was precipitated by adding 800µl of 100% alcohol and 200µl of 7.5M ammonium acetate, and the tubes stored at –20°C for 1 hour. DNA was pelleted by centrifugation at 13500rpm for 10 minutes. The alcohol was aspirated and the pellets washed with

1ml of 70% alcohol and an additional centrifugation step. The pellets were vacuum dried for 30 minutes, then re-suspended in 50µl of TE buffer (pH 8.0) and stored at –

20°C.

DNA was extracted from feather tips following the protocol of Double and Abbott

(1998). A number of other methods were trialed (Chelex, Phenol/Chloroform– isoamyl), but unless the feather had been removed recently from a live bird, failed consistently or produced such a low DNA yield that sequence analysis was impossible.

30 Chapter 2 Genetic Distribution

The number of feathers used in extractions differed. Two to three chest feathers were used where possible. One feather was used for coverts and retrices, and primaries were transversally split down the rachis with half used in the digestion step. These were added to a sterile Eppendorf tube containing 300µl of freshly made digestion buffer (10mM Tris/HCl pH 8.0, 10mM EDTA, 100mM NaCl, 40mM DTT and 2%

SDS) which completely dissolved the tissue. Each tube had 7.5µl of Proteinase K

(20mg/ml) added, then was vortexed briefly and incubated overnight with rotation at

37°C. Final steps were then made using solutions, spin columns and collection tubes supplied with the QIAGEN DNeasy Tissue Kit, with the following changes to their described animal tissue protocol. 200µl of Buffer AL and 210µl of isopropanol were added to each sample and mixed by vortexing. This solution was spun through a

DNeasy spin column at 8000 rpm for 1 minute and the liquid portion discarded. 500µl of Buffer AW1 was applied to the columns, and they were re-spun at 8000 rpm for 1 minute, and the solution discarded. 500µl of Buffer AW2 was added to the columns, and they were given a final spin at 13000 rpm for 1 minute. The spin columns were then transferred to sterile Eppendorf tubes. DNA was eluted by applying 50µl of pre- heated (70°C) Buffer AE to the central membrane of each spin column, followed by incubation at 70°C for 5 minutes prior to centrifuging at 8000 rpm for 1 minute. This was repeated two more times. The incubation step was included as it gives a greater

DNA yield. The final 150µl of eluted DNA template was subjected to a final ethanol precipitation step in anticipation of the extremely low DNA yield from feather tissue.

This step has been described above for blood, however DNA was pelleted at 13500 rpm for 30 minutes and air-dried to reduce the possibility of DNA shearing using a

31 Chapter 2 Genetic Distribution

vacuum drier. Dried pellets were then re-suspended overnight in 25µl of TE buffer

(pH 8.0) and stored at –20°C.

DNA was extracted from skin and other tissue types (heart muscle and liver) in a number of ways. A small portion (~25mg) of “fresh” tissue (frozen dissection, heart muscle or liver) was added to a 1.5ml Eppendorf tube and ground in liquid nitrogen using ‘Kontes’ microtube pestles. The ground tissue was incubated in an extraction buffer (400µl lysis buffer, 20µl 10% SDS, 20µl 1M DTT, 3µl Protienase K) overnight at 37°C with rotation. It was then extracted following the phenol/chloroform-isoamyl alcohol protocol described for blood. All other steps were the same as for blood except that pellets were air-dried and re-suspended in 25 µl of TE buffer (pH 8.0).

Dried skin (museum tissue) was re-hydrated using a 9% NaCl solution for 48 hours prior to extraction. It was then extracted following the above ‘fresh’-tissue protocol or if problematic extracted using the above feather protocol.

It was subsequently discovered that museum skins that were soft and yellowish brown produced very low quantities of DNA or failed altogether. This is possibly due to a high level of fat in the preserved tissue, and has been reported elsewhere by Glenn

(1999).

All extractions were performed including negative controls. DNA was quantified using a spectrophotometer (Quantagene RNA/DNA calculator), so that aliquots of the

32 Chapter 2 Genetic Distribution

raw extractions could be adjusted to allow for a standard concentration of DNA template to be added to PCR reactions regardless of the source tissue.

2.2.2.3 Choice of gene region

Sequence data was prepared using the hyper-variable control region within the mitochondrial genome. In birds the control region has a high level of intra-specific sequence divergence ranging from 12 – 25% across its three component domains

(Baker and Marshall, 1997). As an indirect measure of effective female dispersal and recruitment, and proven utility in the definition of potential conservation management units, this was considered the most appropriate marker for this study (Rand, 1996;

Taylor et al., 2000).

2.2.2.4 Description of isolation of control region and primer design

Preliminary exploration of the control region was made under the assumption that H. leucogaster shared the same gene arrangement as the majority of bird species (Fig.

2.3a). A number of generic primer pairs were trialed to amplify the region bounded by the ND6 gene and tRNAphe at the 5’and 3’ ends of the control region respectively.

The band size, when visualised, was several orders of magnitude smaller than expected for a fragment of avian control region displaying the standard gene order, and it was suspected that H. leucogaster may have the novel gene order described by

Mindell et al. (1998) for a number of other bird taxa including representatives from:

Falconiformes, Cuculiformes, Piciformes and Passeriformes (sub-oscines). (Fig.

2.3b).

33 Chapter 2 Genetic Distribution

a) Standard Avian

Control ND5 Cyt B T P ND6 E F 12s Region b ) Novel Avian CBHL CR-L Site of first CR1-L repeat ND6C-L CBHL-2

Control Region III P ND6 F 12s CytB T III E NC

CR1-H CR2-H CR3-H 12sbi ND6-W

Primer Sequence

CBHL 5' ttc ctc att ctc cta gcc ctc c 3' CBHL-2 5' agt gga ctg cgg gga tat gca c 3' ND6C-L 1 5' ccg aga caa ccc acg cac aag 3' DK12sbi 2 5' aag agc gac ggg cga tgt gt 3' ND6W 5' ctt gtg cgg ggg tgg tct cgt 3' CR-L 5' gtc ata tat atg taa tac ggg c 3’ CR1-L 5' cag ggt cat gtt gca ctg cat g 3' CR1-H 5' cat gca gtg caa cat gac cct g 3' CR2-H 5' gct ggg aat gta gga tct cca c 3' CR3-H 5' ggg ttt gtg aag aat tca tga g 3' 1. Edwards, 1993 2. Palumbi, 1991

Figure 2.3 a) The standard avian gene order compared to b) the novel gene order found in Haliaeetus leucogaster. Primer locations and direction indicated. Primer sequences are listed below. Gene labels are as follows: Cyt B - Cytochrome oxidase B; T - tRNAthr; I, II, III - Domain 1-3 of the Control Region; P - tRNApro; ND6 - NADH dehydrogenase 6; E - tRNAglu; F - tRNAphe; NC - non-protein coding sequence.

34 Chapter 2 Genetic Distribution

Subsequent to this, a complement to the light strand primer originating in the ND6 gene was designed (ND6-W; 5’- CTTGTGCGGGGGTGGTCTCGT-3’). This gene is highly conserved, with the primer sequence between the chicken (Genbank Accession

No. X52392) and the peregrine falcon (Genbank Accession No. AFO69424) differing by only 2 base pairs. The new primer (ND6-W) was the reverse complement of the peregrine falcon.

The novel gene order was confirmed using a generic Cytochrome-Oxidase B primer

(Palumbi et al., 1991) and ND6-W. Following this, a specific primer based on known

H. leucogaster sequence for Cytochrome-Oxidase B was designed (GenBank

Accession No. Z73468). A Long-PCR protocol using the ExpandTM Long Template

PCR System (Boehringer Mannheim) and primer pairs CBHL ( 5’-

TTCCTCATTCTCCTAGCCCTCC-3’) and ND6-W was used to generate a fragment of DNA. This fragment ranged between 2000- 3000bp among different individuals.

The 3000bp fragment was sequenced and internal primer pairs were designed to facilitate the use of standard PCR protocols, and to deal with small fragments of DNA common to museum or degraded samples.

Control region sequence data was allocated to domains I, II and III following the procedure used by Baker and Marshall (1997) and was consistent with the base pair composition reported for each domain. A large number of tandem repeat motifs were found in domain III which contributed strongly to the fragment size variation between individuals reported above. Tandem repeats are frequently subject to a very high rate of evolution, both within the same individual, and between different tissue types

35 Chapter 2 Genetic Distribution

(Lunt et al., 1998). In this data set they were found to be creating a degree of variation inconsistent with the variation based on the nucleotide sequence data external to the repeat units. To avoid attributing artificially high levels of variation to individual sequences, the domain III repeat units were deleted from the analysis.

A large portion of domain II, until the site of the first tandem repeat region, was invariant and was discarded from the final analysis. The final 562 base pair fragment comprised 399 base pairs of domain I and 163 base pairs of domain II and was amplified using the light strand primer CBHL and heavy strand primer CR2-H (5’-

GCTGGGAATGTAGGATCTCCAC-3’). Museum tissue and feather samples were typically degraded in origin, and used primer pairs internal to those above. Sequence products were appended to make the complete fragment. A complete list of primer pairs and their locations in the genome are given in Figure 2.3.

2.2.2.5 PCR conditions

PCR reactions included: 5µl 10x polymerase reaction buffer (Fisher Biotech), 4µl

25mM MgCl2 (Fisher Biotech), 0.2µl Taq DNA Polymerase (5 units/µl – Fisher

Biotech), 10mM each of dATP, dGTP, dCTP, dTTP (Fisher Biotech), 2µl of each primer (10µM), and 60ng of DNA template adjusted to a final volume of 50µl with ddH2O. All PCR reactions contained a negative control. Temperature cycling was as follows: denaturation at 95°C for 5 minutes, then 40 cycles of 95°C denaturing for 1 minute, 57°C annealing for 1 minute, 70.6°C extension for 1 minute followed by a final extension of 7 minutes at 70.6°C. Internal primer combinations used the same

36 Chapter 2 Genetic Distribution

temperature profiles but required different optimum annealing temperatures. They were as follows: CBHL * CR1-H, 66.5°C; CR-L * CR2-H, 65°C.

PCR product was visualised and compared to a known marker on a 1.6% agarose gel containing ethidium bromide. Product was cleaned using a QIAGEN Qiaquick PCR purification kit as per the manufacturers instructions. 10µl Sequence reactions were prepared using 4µl dye terminator mix (Perkin Elmer), 3.2pmol of primer and 30ng of

DNA template. Reactions were adjusted to their final volume using ddH2O. These were subject to: 25 cycles of 96°C for 30 seconds, 50°C for 15 seconds and 60°C for

4 minutes. DNA was cleaned according to manufacturers instructions and sequenced on an Applied Biosystems 377 automated sequencer. Sequence data were checked for a subset of individuals using forward and reverse primers.

2.2.3 Data analysis

Sequences were aligned using the program BIOEDIT (Hall, 1999). Genetic variation within each region was summarized using haplotype and nucleotide diversity statistics. Haplotype diversity was calculated in the package REAP (Version 4.0;

McElroy et al., 1991) using the equation of Nei (1987). Nucleotide diversity among haplotypes within regions was calculated in ARLEQUIN (Version 2; Schneider et al.,

2000). Evolutionary relationships among haplotypes were examined as a network of mutational differences using TCS (Version 1.13; Clement et al., 2001) where alignment gaps were considered as a fifth state.

37 Chapter 2 Genetic Distribution

Population subdivision within and among regions, and within and among pooled regions was calculated using Analysis of Molecular Variance (AMOVA). Pooled regions (groups) are defined in Table 2.6. AMOVA is considered analogous to

Wright’s F-Statistics (Wright, 1951), and considers a less rigid set of assumptions more applicable to the analysis of single locus molecular data that may be subject to a range of different demographic scenarios (Wright, 1951; Weir and Cockerham, 1984;

Slatkin and Maddison, 1989; Excoffier et al., 1992; Neigel, 2002). Standard FST values were calculated at the continental level and including all sample regions. Φ statistics, including pairwise comparisons among geographic regions, were generated considering both haplotype frequency and evolutionary divergence based on the distance method of Tamura and Nei (1993) with a gamma value of 0.09, which accounts for differing nucleotide substitution rates at different sites within the overall fragment. The gamma value was calculated based on each unique haplotype in the program PUZZLE (Version 4.0.2; Strimmer and von Haeseler, 1996). Φ statistics and associated variance components were tested for significance using 1600 random permutations. All AMOVA procedures were calculated using the ARLEQUIN package (Schneider et al., 2000). The effective female gene flow (Nefmf) between regional groupings was calculated using the equation Nefmf = ½(1/ΦST – 1).

Mantel tests using both coastline and straightline distance between regions and

Slatkin’s linearised ΦST were used to investigate whether there was a relationship between geographic proximity and gene flow. Geographic distance was calculated between regional centroids in the GIS package ARCVIEW 3.2 (ESRI, 1996). To test for selective neutrality, sequences were tested using Tajima’s D statistic (Tajima,

38 Chapter 2 Genetic Distribution

1989). This had the added benefit of distinguishing organismal and molecular demographies where the derived statistic suggests two distinct scenarios (1) balancing selection or admixture (positive (+) Tajima’s D), or (2) a selective sweep or recent population bottleneck (negative (-) Tajima’s D) (Rand, 1996). Tajima’s D was calculated for each variable geographic region and for the entire sample using

ARLEQUIN.

To examine further the hypothesis that the Australian population was the result of a recent population expansion from Southeast Asia, the observed distribution of pairwise nucleotide differences between all Australian haplotypes was compared to that expected under the sudden expansion model of Rogers and Harpending (1992).

The mismatch distribution between observed and expected differences was calculated using the least-squares method of Schneider and Excoffier (1999), which is thought to predict more accurately the age of expansion (Schneider et al., 2000). Unless otherwise stated, all statistics were generated in ARLEQUIN. The extent to which the observed distribution differed to that expected was tested using a coalescent approach adapted from Hudson (1990) based on the sum of square deviation (SSD) between the observed and expected mismatch. The raggedness (r) value generated in ARLEQUIN was tested for significance by a coalescent procedure using DNASP v3.50 (Rozas and

Rozas, 1999).

An average rate of 20% divergence per million years has been calculated for control region I in bird species (Baker and Marshall, 1977). Here I used a divergence rate of

20.8% per million years based on the snow goose Chen caerulescens (Quinn, 1992).

39 Chapter 2 Genetic Distribution

Wenink et al. (1996) have estimated a divergence rate of 8.8% for control-region II.

Following the suggested method of Wenink et al. (1996) an approximate sequence divergence rate was estimated as follows. Two-thirds of the assayed region is within control-region I. By using two-thirds of the divergence from control-region I, and a third of the divergence from control-region II, an approximate rate of 16.6% divergence per million years was applied to the entire fragment. Time since expansion was calculated using the formula τ = 2ut; where τ = units of mutational time and u = the mutation rate over the fragment assayed.

Nested clade analysis (Templeton, 1998) was used to differentiate between signatures of historical and current gene flow among the Australian regions. Insufficient sampling of intermediate sites separating Singapore from Australia precluded the inclusion of overseas data within this analysis. It is not known whether long distance dispersal in H. leucogaster is restricted to coastal or riparian movement in favour of more direct overland routes. Given known ecology and physical sighting records

(Blakers et al., 1984), it is reasonable to assume dispersal occurs along coastal routes.

Thus, a matrix of pairwise differences of geographic coastal distance between regional centroids was used for this analysis as recommended in Posada et al. (2000).

The same matrix of distances applied in the Mantel tests was used. Clade associations were constructed using the estimated cladogram produced in TCS following the nesting rules of Templeton et al. (1987) and Templeton and Sing (1993).

Additionally, an unrooted neighbour joining tree using the Tamura and Nei (1993) gamma distance method (α =0.09) was generated in MEGA 2.0 (Kumar et al., 1993)

(not shown). This oriented two out of three of the Singapore samples to the basal

40 Chapter 2 Genetic Distribution

node (bootstrap value = 67%) effectively placing them as an outgroup (Page and

Holmes, 1998). This information, in combination with the a priori assumption that

Australia was colonised from Southeast Asia, was used to designate clade 1-1 as an interior rather than a tip. The geographic relationship within and between clades, at each level within the cladogram, was tested for significance at the 5% level in

GEODIS 2.0 (Posada et al., 2000) using 5000 re-samples, and interpreted using the updated inference key (20 December 2000) for nested haplotype tree analysis. This key originally appeared in Templeton et al.(1995).

2.3 Results

2.3.1 Variation within the fragment

Mitochondrial DNA (mtDNA) was successfully isolated from all sample types.

Sequence data was obtained for 128 individuals covering the Australian range and including individuals from Singapore (N=3) and Papua New Guinea (N=4).

An additional two repeat units were found in domain I of the control region. In

Haplotype 8 (HL8), the second nine-base ‘C’ repeat was missing (Table 2.2). It was unclear if this indel represented one or more mutational events (Estoup et al., 2002).

In all other samples this repeat was phylogenetically non-informative, and it was removed from the analysis on the basis that it could impose an artificially high level of divergence for this haplotype. This reduced the final fragment size from 562 to

553 bp including two alignment gaps.

41 Chapter 2 Genetic Distribution

A total of 15 polymorphic sites described 15 haplotypes including three endemic to

Singapore (GenBank Accession No’s: AY225291-AY225305). The Singapore haplotypes differed from the remaining haplotypes by only three mutations and from each other by single base pair changes. Notably, the Singapore haplotypes (HL13,

HL14 and HL15) lacked the alignment gaps found in domain I, and the nucleotide

CR Domain I II

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 3 3 4 4 5 1 1 3 5 2 7 8 8 8 8 8 8 8 8 8 8 9 9 9 9 9 9 9 9 9 1 1 3 4 6 1 7 8 9 5 9 9 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 0 4 3 0 5 3 HL1 G TC C T A C CCCCCCCCTTCCCCCC C C ~ C C T TA C HL2 ...... ~ ...... C . . C . . . HL3 . . . T ...... ~ ...... ~ ...... HL4 . . T ...... ~ ...... C ...... HL5 C ...... ~ ...... C ...... HL6 . . T T ...... ~ ...... C ...... HL7 ...... ~ ...... C . . . . . T HL8 ...... C ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ...... HL9 C A ...... ~ ...... ~ ...... HL10 ...... ~ ...... C A . . . . . HL11 . . . . A ...... ~ ...... C ...... HL12 . . . . . G ...... HL13 ...... C . . HL14 ...... T . . . . HL15 ...... C T .

42 Chapter 2 Genetic Distribution

mutations differentiating HL13 and HL15 occurred within the more conserved

domain II. The remaining 12 haplotypes were found in Papua New Guinea and

Australia (Fig. 2.4; Table 2.5). Papua New Guinea showed two haplotypes, HL1 and

HL2. HL1 was shared among all Australian regions. However, Tasmania was the only

other region to display HL2. The fragment contained eight transition changes (C↔T,

A↔G) and four transversions (G↔C, C↔A, T↔A). A total of five sites were

considered parsimoniously informative.

Region 8 (N= 4) Region7 (N=3)

Haplotype 1 Region 4 2 3 (N=13) 4 5 6 7 8 9 10 11 12 13 14 15 Region 3 (N=20)

Region 5 (N=10) Region 2 (N=22)

Region 6 (N=17)

Region 1 (N=39)

Figure 2.4 Frequencies of individual haplotypes from each region assayed. Where: Region 1 = Tasmania; Region 2 = South Eastern Australia; Region 3 = Central & Southeast Queensland; Region 4 = Northern Australia; Region 5 = Western Australia; Region 6 = South Australia; Region 7 = Singapore; Region 8 = Papua New Guinea.

43 Chapter 2 Genetic Distribution

Both nucleotide and haplotype diversity were low (Table 2.3). Haplotype diversity ranged from zero in Southeast Australia to 0.574 ± 0.219 in Central and Southeast

Queensland, with an average diversity within Australia of 0.314 ± 0.055. Southeast

Australia was the only region fixed for a single haplotype, and the only region to show significantly different pairwise ΦST values when compared with other regions

(Table 2.4). The low haplotype diversity values are attributable to the imbalance between haplotype frequencies. The most common haplotype (HL1) occurred with an average frequency of 81% across regions with the remaining haplotypes occurring at relatively low frequencies (Table 2.5).

2.3.2 Evidence for genetic structuring within Australia

There appears to be some restriction to gene flow given the number of haplotypes restricted to individual regions (Fig. 2.5). However, this was not supported by results of the AMOVA analysis.

I was particularly interested in evidence to indicate a restriction of movement between the west and east of Australia, or between Tasmania and the mainland.

Results failed to provide any significant level of subdivision between imposed groupings at any of the geographic scales, suggesting these groupings were artificial

(Excoffier et al., 1992) (Table 2.6). Overall, the percentage of variation both within regions and between groupings was very low (Table 2.6). A number of significant

ΦSC values indicated some degree of variation within groups (pooled regions; see

Table 2.6), this was probably a result of the variation in the frequency of the most common haplotype among regions within groups.

44 Chapter 2 Genetic Distribution

Table 2.3 Mitochondrial DNA control region haplotype and nucleotide diversity within regions in Australia and outside Australia including neutrality tests.

Region No. of Haplotype diversity (†) Nucleotide diversity Tajimas’s D(Ψ) haplotypes Within Australia (1) Tasmania 5 0.2834 ± 0.09216 0.000565 ± 0.00029 -1.51114 (0.05922) (2) Southeast 1 0 0 - Australia (3) Central & 6 0.5737 ± 0.2185 0.001421 ± 0.00058 -0.35167 (0.38397) Southeast Queensland (4) Northern 2 0.1538 ± 0.1261 0.000294 ± 0.00034 -1.14915 (0.13689) Australia (5) Western 4 0.5333 ± 0.18007 0.001129 ± 0.00109 -0.88367 (0.21514) Australia (6) South Australia 4 0.4191 ± 0.14129 0.001017 ± 0.000975 -0.26239 (0.41707) Within 12 0.3138 ± 0.05468 0.000695 ± 0.000732 -1.84467 (0.019) Australia - Total Outside Australia (7) Singapore 3 1.0 ± 0.27217 0.003869 ± 0.00208 22284.64 (8) Papua New 2 0.5 ± 0.26517 0.000944 ± 0.00081 -0.61237 (0.4138) Guinea Total 15 0.3497 ± 0.05447 0.000806 ± 0.0008 -2.01194 (0.00984) Values after Haplotype and Nucleotide diversity are standard errors. † 2 Haplotype Diversity = n(1- ∑x i)/(n-1), where x is the frequency of the ith haplotype and n is the number of individuals within each region. Ψ Figures in brackets are the probability of D random < D obs Shaded cells are significant at P<<0.05

45 Chapter 2 Genetic Distribution

Table 2.4 Pairwise ΦST between each geographic region. Regions are 1) Tasmania 2) Southeast Australia 3) Central and Southeast Queensland 4) Northern Australia 5) Western Australia 6) South Australia. Values in bold are significant at P<0.05

1 2 3 4 5 6 1 0.000 2 0.003 0.000 3 0.005 0.057 0.000 4 -0.0002 0.043 0.023 0.000 5 0.029 0.079 -0.002 0.012 0.000 6 0.013 0.126 -0.005 0.065 0.059 0.000

At the continental scale the degree of variation between all regions approached significance (P=0.07), but overall variation was very low (2.32%) due to the dominance of HL1 in all regions.

46 Chapter 2 Genetic Distribution

Table 2.5 Frequencies of individual haplotypes from each region assayed.

Region (1) Tasmania (2) Southeast (3) Central & (4) Northern (5) Western (6) South (7) Singapore (8) Papua Australia Southeast Qld Australia Australia Australia New Guinea n 39 22 20 13 10 17 3 4 Haplotype

HL1 0.846 1 0.65 0.923 0.7 0.765 0.75 HL2 0.0256 0.25 HL3 0.0256 0.0588 HL4 0.0769 0.15 0.118 HL5 0.0256 0.05 0.1 HL6 0.0588 HL7 0.1 HL8 0.1 HL9 0.05 HL10 0.05 HL11 0.05 HL12 0.0769 HL13 0.33 HL14 0.33 HL15 0.33

Table 2.6 Φ statistics and the hierarchical partitioning of genetic variation within and between pooled regions calculated from AMOVA using the distance method of Tamura & Nei (1993) and a gamma value of 0.09, where: ΦSC measures the proportion of genetic variation among regions within groupings; ΦCT measures the proportion of variation between groups; and, ΦST measures the proportion of genetic variation between all regions. Standard FST calculations are given in brackets. The theoretical movement of females between groups per generation is also included.

Φ Statistics ∝ Σ Comparison Groupings ΦSC Φ CT ΦST Nefmf North vs South 0.026 -0.005 0.021 28.3

NW vs SE 0.019 0.010 0.029 16.7

Tas vs All Regions 0.045* -0.043 0.004 124.5

Tas vs East Coast 0.065* -0.07 0.003 166.2

Tas vs East Coast & South Australia 0.053 -0.052 0.004 124.5

Tas vs East Coast & South Australia 0.05* -0.047 0.003 166.2 & Northern Territory WA vs All Other Regions 0.018 0.026 0.044Ψ 10.9

Among All Regions Within 0.023# 21.2 Australia (0.036*) Among All Regions Including PNG 0.095** 4.8 and Singapore (0.098*)

∝ Where artificial groupings are: North = Region 4, 5, 3; South = Region 1, 2, 6; North West = Region 4, 5; South East = Region 1, 2, 3, 6; East Coast = 2, 3. Figures in brackets show total % variation for statistic. Ψ # Σ * = P < 0.05, ** = P < 0.001; P = 0.0695; P = 0.07; Nefmf = ½ ( 1/ΦST – 1). 49 Chapter 2 Genetic Distribution

There was no effect of isolation by distance based on the Mantel tests using either coastline (r = -0.086; P = 0.627) or straightline (r = -0.309; P = 0.878) distance between regions (Fig. 2.6). Nested clade analysis supports these conclusions; see below.

Figure 2.6 Scattergram for coastal and straightline distance against Slatkin’s linearised ΦST using Ln + 1 transformed values. Distances between the mid-point of each region were calculated using ARCVIEW 3.2. P values indicate the level of significance of the Mantels correlation statistic (r) using 10 000 permutations.

50 Chapter 2 Genetic Distribution

2.3.3 Evidence for a historical bottleneck and population expansion

Within regions there were no significant deviations from neutral expectations, although the calculated Tajima’s D value was consistently negative for each geographic region. Both at the continental scale, and when including Papua New

Guinea and Singapore, highly significant negative values were obtained suggesting a recent population bottleneck (Table 2.3). Also, there was no significant difference between the observed and expected distribution under the expansion model

(SSD=0.07, P=0.084) (Fig. 2.7).

6000 Observed

5000 Expected

4000 y y θ = 0 c c 0 n n θ =0.352 ue ue 3000 1 eq eq r r τ = 3.0 F F 2000 r = 0.5108 t = ~160 000 yrs 1000

0 0123456 No. of pairwise differences

Figure 2.7 The observed and expected distribution of pairwise differences used to test for deviation from the sudden expansion model (Rogers & Harpending, 1992). Statistics used to infer time since expansion and population size before and after expansion are also shown.

51 Chapter 2 Genetic Distribution

Similarly, the observed distribution had an extremely small probability of being more ragged than expected under the expansion model (r=0.51, P=0.996). That is, expanding population distributions are characteristically unimodal, while those under equilibrium conditions are ragged exhibiting many peaks (Rogers and Harpending,

1992). The calculated tau value (τ=3.0; 95% CI: 0.395-4.25) and θ parameters (θ0=0;

θ1=0.352) are consistent with a population that has expanded from a very small initial population approximately 160 000 years BP (95% CI; 21 000 – 228 000 yrs bp), with the extant population retaining a relatively small effective population size (Rogers and Harpending, 1992).

Four levels were resolved in the nested clade analysis (Fig. 2.8). Five 1-step clades were divided into two 2-step clades, separated by a single unsampled haplotype.

Clade 1-1 and the total cladogram were the only clades in which a significant association was found between haplotype and location (Table 2.7). The cosmopolitan distribution of clade 1-1 is shown in Figure 2.9. The inference key suggests this is due to a contiguous range expansion following the initial colonisation (Templeton, 1998)

(Table 2.7).

52 Chapter 2 Genetic Distribution

Figure 2.8 Nested clade diagram derived from the original cladogram (Fig. 2.5). The Singapore haplotypes have not been included. Individual haplotypes correspond to 0- step clades. 1-step clades are enclosed within dotted lines, 2-step clades within the solid lines. The total cladogram is enclosed within a dashed polygon.

53 Chapter 2 Genetic Distribution

Table 2.7 Summary table showing the only clades with a significant permutational chi-square probability following 5000 re-samples where: the clade distance (Dc) measures the geographical range of the nominated clade; the nested clade distance (Dn) measures the geographical relationship between the nominated clade and its nearest evolutionary sister clade; and I-T describes the average difference between interior vs tip clades for both distance measures (Templeton, 1998). Significantly small distance measures at the 5% level are in bold and indicated by a superscript S. The original nesting design is given in Fig. 2.8. The final biological conclusion was obtained by following the stated path in the inference key (Templeton et al., 1995).

Nesting Level Location Haplotype Dc Dn Inference Key path and conclusion

1-1 Interior HL1 4317.6S 4379.9S Tip HL8 0 7545.9 1→ 2 → 11b → 12 Tip HL12 0 6638.1 Contiguous range expansion I-T 4317.6S -2712.1S

Total 2-1 4417.7 4391.4 1 → Extensive Cladogram dispersal/ Panmixia/ 2-2 5165.2 4356.1 or small sample size/ inadequate geographical sampling

54 Chapter 2 Genetic Distribution

2.4 Discussion

Theoretically, a population that has expanded from a genetic bottleneck will show evidence of: a low historical Ne with low haplotype and nucleotide diversities (Glenn,

1999; Mila et al., 2000); a star-like phylogeny of haplotypes with very low levels of population subdivision; and, a unimodal distribution of pairwise differences among haplotypes (Zink, 1997; Mila et al., 2000). The haplotype and nucleotide diversity values in this study are considered low when compared with other bird species

55 Chapter 2 Genetic Distribution

(Vallianatos et al., 2002), but similar to species which have suffered a severe historical bottleneck , for example the Knot (Calidris canutus) (Baker and Marshall,

1997). The Tajima’s D values, estimated cladogram, and mismatch analysis suggests that the Australian population has expanded from a historical bottleneck, although without corroborative data from a second marker it is not possible to discount the effect of a selective sweep at some stage in the past (Rand, 1996; Ray et al., 2003).

Additionally, when viewing the graph of pairwise differences from the mismatch analysis (Fig. 2.7), the small peak at 3 is symptomatic of either population structure or an additional earlier bottleneck (Zink, 1997). A similar result was obtained by Zink

(1997) in the fox sparrow (Passerella iliaca), although there was clear evidence of population structure for this species (Zink, 1994). Given the apparent lack of structure between populations of H. leucogaster, an earlier bottleneck event is more likely.

According to Templeton (1998), for the inference key to impute a contiguous range expansion, the expanding population must carry a subset of the variation from the ancestral population (including at least one tip clade) to the expanded area. There must also have been sufficient restriction to gene flow among ensuing generations for the mutational process to create new tip clades (Templeton, 1998). In the case of an extreme founder event, or extremely recent colonisation, where there has been insufficient time for new mutations to arise, a range expansion pattern would not be found (Templeton, 1998). The close cladistic relationship between the most common haplotype (HL1) to those found in Singapore, and this haplotype’s ‘blanket’ distribution within Australia, suggest it as the most likely ancestral haplotype given the current sample (Fig. 2.5). HL8 and HL12, both tip haplotypes, belong to clade1-1

56 Chapter 2 Genetic Distribution

of the nested analysis and theoretically form part of the original founding population.

The remaining haplotypes, by virtue of their derived position in the cladogram, have probably arisen in situ. The mismatch analysis suggests a very small initial colonising population approximately 160 000 years ago, or within the last 32 000 generations

(based on an approximate generation time of five years; Marchant and Higgins,

1993), during which there has been sufficient time for these new haplotypes to appear. The shared distribution of HL3, HL4 and HL5 is suggestive of current gene flow (Zink, 1997), but may also indicate ancestral retention following historically more recent range expansions within Australia.

Both the AMOVA analysis and estimated Nefmf values (Table 2.6) indicate strong current connectivity between regions. Within Australia, mark-recapture records are limited, but long-distance dispersal capability in H. leucogaster is substantiated by an

Australian Bird and Bat Banding Scheme record of a juvenile that moved 3000 km from its natal territory in South Australia to Fraser Island in Queensland (Marchant and Higgins, 1993). Using a conventional FST approach (Wright, 1951), the movement of only a single individual per generation (Nm = 1) between populations yields an FST of 0.2 and is sufficient to prevent divergence to fixation (Avise, 1994;

Zink et al., 2000). This value is many orders of magnitude larger than the calculated

ΦST values within this study or the standard FST values attained at the continental scale and across the whole study (Table 2.6). Given the large body size and proven dispersal potential of the species, it was not surprising that there would be evidence of strong gene flow. However, there are a couple of anomalies which suggest the potential for a higher level of population structure than indicated by either the

AMOVA analysis or the estimated Nefmf values. Firstly, high levels of inferred

57 Chapter 2 Genetic Distribution

current gene flow are inconsistent with the large number of rare haplotypes that appear unique to individual regions, although Excoffier (2004) states that an excess of rare mutations is expected following a demographic expansion. Secondly, the

Southeast is the only region fixed for a single haplotype.

A number of explanations are possible for the non-sharing of haplotypes between the north-west and east of Australia. First, sample sizes in the west and north of the continent are an order of magnitude smaller than in the eastern regions. Small sample size or inadequate geographic sampling were alternate findings of the nested clade analysis at the level of the total cladogram. A larger sample size may see the occurrence of eastern haplotypes in the north and west. However, this explanation does not necessarily account for the lack of HL7, HL8 and HL12 in the eastern regions where sample sizes are much larger. Second, low frequency haplotypes are only expected to attain high frequency within a population where the effective movement between demes is low (Avise, 1994).

Given the relatively small number of generations since colonisation, it is possible that there has been insufficient time to reach equilibrium between gene flow and genetic drift (Wright, 1951). AMOVA analysis is based on the assumption that populations are in equilibrium (Excoffier et al., 1992). Where populations have not yet reached equilibrium, an apparent lack of structure may be interpreted as current gene flow, as suggested by the AMOVA, but may equally describe a historical pattern of haplotype association (Greenberg et al., 1998), and an inflated value of Nemf that is biologically misleading (Templeton, 1998).

58 Chapter 2 Genetic Distribution

Similarly, if HL7, HL8 and HL12 are present within the east of Australia, there may not have been enough time (Crandall and Templeton, 1993) for them to become established at high enough frequency within other regions to be discernible withthe sample numbers used in this study. This may also explain the sharing of HL2 between

Papua New Guinea and Tasmania. This haplotype may exist at extremely low frequency among other regions, and was only sampled within Tasmania due to the comparatively large sample size in this region. Alternatively, it may have been lost from the geographic intermediates over time due to its rarity (Futuyma, 1986).

The apparent homogeneity and significant pairwise ΦST comparisons in the Southeast are difficult to explain. This population may represent a recolonisation following local extinction at some stage in the past, or could reflect a more recent response to habitat change (see Chapter 3) effectively producing a localised bottleneck.

Specifically, the Victorian population is currently considered restricted with only 100 pairs estimated as remaining (Clunie, 1994).

There is insufficient evidence to suggest the division of the population to different units for management purposes based on the theoretical definition of the

‘evolutionary significant unit’ (ESU) (Moritz, 1994). To satisfy this definition, mitochondrial haplotypes must be reciprocally monophyletic among regions based on haplotype variation. Given that the continental population is the result of a recent expansion, this result was not unexpected (Zink et al., 2000). Typically, bird species contain two or fewer ESU’s, even where species have been sampled across their entire range (Avise and Walker, 1998). However, given the suggested colonisation history of the species, this finding requires verification with a second marker to rule

59 Chapter 2 Genetic Distribution

out the possibility of additional structure. Additionally, this study describes female population structure only. A different pattern of structure may be obtained using a recombinant DNA marker.

An a priori assumption of this study was that H. leucogaster colonised Australia as part of a range expansion from Southeast Asia, assuming the western extreme of the range to be the southern coast of India and the eastern extreme the Australian continent. From the neighbour-joining tree, the basal relationship of two of the three

Singapore samples relative to the Australian haplotypes endorses this view, and in combination with low haplotype diversity across the entire study suggests the

Australian haplotypes hold a derived position within the species’ total range network.

However, a maximum of only two mutational changes separate the Singapore haplotypes from HL1. This suggests the overall evolutionary tree has the potential to be quite shallow and that further sampling within Southeast Asia would likely reveal the presence of HL1. Whilst it was impossible to test these hypotheses using the current sample, a number of suggestions can be made regarding the mechanisms of expansion into Australia.

2.4.1 Reconstruction of colonisation history

Considerable emphasis has been placed on the role of Pleistocene climate change in the genetic diversification of bird species (Avise and Walker, 1998). 160 000 years

BP, during the late Pleistocene, sea-levels were at their lowest, approximately 150m below current levels, which effectively joined the islands of the Sunda shelf to the

Asian mainland and exposed the Sahul shelf (Chappell, 1983; Hewitt, 2000; Karns et

60 Chapter 2 Genetic Distribution

al., 2000). Although deep-water barriers separated the two shelves (Karns et al.,

2000), distances between islands were relatively small and unlikely to pose a significant barrier to the movement of birds or the rapid expansion of H. leucogaster through Malaysia to Australia (Keast, 1981a; McManus, 1985; Karns et al., 2000).

The timing of low sea-levels is remarkably consistent with the genetically based estimate of 160 000 years since expansion into Australia. Papua New Guinea and

Tasmania were also repeatedly joined to the Australian mainland during this time

(Kikkawa et al., 1981; Gopurenko and Hughes, 2002), providing opportunities for panmixia between Australia and Papua New Guinea. Both Timor and New Guinea are known to have functioned as an entry route to Australia for other bird species

(Keast, 1981a; McManus, 1985; Ericson et al., 2002). Until only 8000 years BP, exposure of the Sahul shelf reduced the distance between Timor and Australia to less than 100 kilometers (Keast, 1981a; Karns et al., 2000). The retention of a Clade 1-1 haplotype in Western Australia is suggestive of colonisation via this route. Similar patterns of post-glacial expansion have been described for northern hemisphere species in which leading edge colonisers rapidly fill new territory producing populations with reduced genetic heterozygosity (Hewitt, 2000).

It is clear from the analysis that at least clade 1-1 haplotypes colonised Australia as part of a contiguous range expansion, and that the current population has expanded from a relatively small number of individuals. The extent to which the current association of haplotypes may be due to contemporary gene flow or historical haplotype retention is unclear, notwithstanding the fact that mark-recapture data suggest this species is capable of very long distance movement. Certainly the power of nested clade analysis to discriminate between current and historical processes has

61 Chapter 2 Genetic Distribution

been shown to reduce where geographic sampling may be inadequate or genetic resolution is lacking (Templeton, 2004).

Based on the current marker, it appears that conservation management is appropriate at the continental level. However, given the suggestion that population expansion has been relatively recent, the conclusion of continent-wide management based on

AMOVA may be premature. Additional geographic sampling would be useful to clarify issues surrounding the degree of current gene flow. Furthermore, given the small physical distances between Australia, Papua New Guinea and Southeast Asia, it would be valuable to explore the existence of current genetic exchange or divergence between these geographic areas.

Conclusions from this analysis provide a basis for understanding patterns of genetic distribution, genetic diversity and population extent and therefore provide a useful baseline against which to test future change in population structure (sensu Villianatos et al., 2002). However, the data say little about population movements at ecological timescales. This is the focus of the next chapter.

62 Chapter 3 Distributional Variation

Chapter 3 – Variation in the distribution of the White-bellied

Sea-Eagle across Australia at ecological time-scales

3.1 Introduction

At the continental scale in Australia there is a general lack of long-term data available for time-series comparison of distributions of animal species. However, the

Australian Bird Atlas data provide a unique opportunity to explore distribution and quantify bird abundance (Garnett, 2001; Turner et al., 2001). To date, research using the Australian Bird Atlas data has mainly concerned seasonal movement (Clarke et al., 1999; Griffioen and Clarke, 2002), the identification of decline amongst passerine species (Peters, 1979; Franklin, 1999; Reid, 2000; Garnett, 2001), and more recently an exploration of broad distributional trends (Garnett et al., 2002; Barrett et al.,

2003). Bird Atlas data have been used extensively in the northern hemisphere to explore questions related to population status and trends in abundance (James et al.,

1996; Kirk and Hyslop, 1998), the estimation of absolute population size (Robertson et al., 1995), the interplay between habitat and the distribution and abundance of species (Flather and Sauer, 1996), and the effect of disturbance regimes

(e.g.Chamberlain et al., 2000).

A discussion of abundance/occupancy relationships and the utility of atlas data sets is given in Chapter 1.2.2. In this chapter I analyse the Australian Bird Atlas Data to identify the extent and pattern of change in range or density of H. leucogaster in

Australia over differing spatial and temporal scales between 1901 and 2001.

63 Chapter 3 Distributional Variation

Specifically I was interested in 1) identifying trends of persistent decline at either the regional or continental scale, 2) movement into previously unoccupied habitat, 3) changes in coastal distributions, and 4) exploring contemporary factors that may be contributing to change in abundance or range extent.

3.2 Methods

3.2.1 Source of the Data

Records of bird sightings were obtained from the Australian Bird Atlas data administered by Birds Australia. Distribution records were available over five time periods grouped by Birds Australia into three distinct ‘Atlas Periods’: The Historical

Atlas (1901-1976), The Field Atlas (1977-1981; referred to here as Atlas 1) and the

New Atlas (1998-2001; referred to here as Atlas 2).

The Historical Atlas records were compiled from observers’ notebooks, published records, museum and private collections (Blakers et al., 1984). Atlas 1 and 2 data were collected during a systematic continental sampling regime involving thousands of volunteer bird watchers (Blakers et al., 1984; Clarke et al., 1999). The atlases used different survey methodology and are not strictly comparable. The basic measurement unit for each Atlas Period was the record sheet. Each sheet contained a list showing which bird species were recorded during a given survey undertaken at a specified time and geographic location. Over 100 species could be recorded by a single observer on any one sheet. All species recorded on a sheet are linked to a single latitude/longitude reference point. Incidental sightings that were not part of a survey

64 Chapter 3 Distributional Variation

were recorded on “incidental sightings sheets” that could contain up to 10 different species records from disparate latitude/longitude reference points. Incidental records were used in both the Historical Atlas and Atlas 1. To ensure records on these sheets could still be identified uniquely, Birds Australia allocated ‘Link Numbers’ to records not sharing the same physical location. Thus, both ‘Sheet’ and ‘Link’ numbers were used by Birds Australia to identify unique sheet records. All incidental sightings were recorded as independent sheet records in Atlas 2, circumventing this issue.

For the present study, three sets of data were obtained at the continental scale:

1. The total number of record sheets per 1° grid block for each Atlas Period.

2. All H. leucogaster records for each Atlas Period.

3. All records for 21 other selected aquatic species for each Atlas Period. These

included marine and freshwater species which had similar habitat (or other)

requirements, or were prey items for H. leucogaster. They were chosen to give

adequate spatial coverage of the entire continent both within and outside the

observed range of H. leucogaster. We considered that an observer recording

any of the nominated species would have a reasonable probability of seeing H.

leucogaster if it was present during the same observation period. Spatially co-

occurring species produced multiple sheet records. These were rationalised to

unique records, each representing a single record sheet on which one or more

of the target species was recorded. The aquatic species (512 593 individual

records) were: Little Penguin Eudyptula minor, Great Cormorant

Phalacrocorax carbo, Little Black Cormorant Phalacrocorax sulcirostris,

Black-faced Cormorant Leucocarbo fuscescens, Pied Cormorant

Phalacrocorax varius , Little Pied Cormorant Phalacrocorax melanoleucos,

65 Chapter 3 Distributional Variation

Darter Anhinga melanogaster, Australasian Gannet Morus serrator,

Australian Pelican Pelicanus conspicillatus, Whiskered Tern Chilidonias

hybrida, Crested Tern Sterna bergii, Silver Gull Larus novaehollandiae,

Pacific Gull Larus pacificus, Australian White Ibis Threskiornis aethiopica,

Straw-necked Ibis Threskiornis spinicollis, Black Swan Cygnus atratus,

Plumed Whistling Duck Dencrocygna eytoni, Pacific Black Duck Anus

superciliosa, Brahminy Kite Milvus indus, Whistling Kite Milvus sphenurus,

Osprey Pandion haliaetus.

Data used for this study were extracted from the original databases and manipulated within a geographic information system (GIS; ESRI 1996). The information analysed included: Atlas Period, species name, the sheet or link number, and the latitude and longitude reference point for each survey. Retaining the sheet or link number allowed all future information to be cross-referenced to the base data if required.

3.2.2 Treatment of the Data

A total of 14 413 records of H. leucogaster were available across all Atlas Periods

(Fig. 3.1a). Link numbers were used to identify records within the Historical Atlas.

Standard sheet numbers were used in Atlas 1 and Atlas 2. A small number of incidental surveys affected the Atlas 1 data. These were deleted from the analysis, removing a total of N=11 H. leucogaster sightings and N=46 aquatic sheet records.

These surveys were within heavily occupied blocks, and were considered to have negligible impact on the final analysis.

66 Chapter 3 Distributional Variation

Using the GIS, a 1° grid was laid across the continent. Each 1° block (approximately

100 km2) was given a unique numeric identifier. A total of 783 blocks were common among all three Atlas Periods. For regional comparisons, blocks were assigned to regions with a similar spatial extent to those applied in the genetic analysis (Chapter

2) (Fig. 3.1b). These regions corresponded loosely with the major fluvial divisions of

Australia (Fig 2.1b), but were principally chosen to test for structural partitioning between areas that could potentially be divided by natural barriers, where known physical breaks occur in the species range, or in keeping with accepted biogeographic boundaries (Thackway and Cresswell, 1995).

Only blocks containing 30 or more sheets (of any species) in all three Atlas Periods were used in the analyses (N=317) (Fig. 3.1b). This ensured that blocks in which H. leucogaster were recorded as ‘absent’ reflected true absence, rather than insufficient sampling effort, and that comparisons across the three periods were valid. The number of blocks analysed for each region were: Central Australia 32, Central and

Southeast Australia 39, North Queensland 26, Northwest Australia 22, Central and

Southwest Western Australia 55, South Australia 26, Southeast Australia 102 and

Tasmania 15. Within the Historical Atlas, a total of 113 408 sheet records were available. Of these, 97 940 sheet records fell within blocks containing 30 or more sheets, of which 16 610 sheets met the criteria defined for use in the analysis. These values were: 89 942, 77 780, 48 295 and 174 739, 168 220, 68 377 in Atlas 1 and

Atlas 2 respectively. There were 1 782, 4 888 and 6 480 H. leucogaster observations from the Historical, Atlas 1 and Atlas 2 data-sets within the selected blocks.

67 Chapter 3 Distributional Variation

# a) H. leucogaster observation across ## # ############### # ########################### ######################### #### ####### all Atlas Periods (N=14413) ########## ############### # # ### ## # # # # #### ### # # ####### # # ## ############################################################################################################################################################################################################################################################## ################################# ########## # ##### ##################################################################### ### # ## ## ########################## ####################### # ######################### ##### ### ################ ############ # # # ####### # # ######################## ######## ######################### ####### # ### # # ##### # ## # ## ##### ## ### #### ########## # ## ##### ##### ##### ######### ###### # # ##################### ###### # ######## #### ####################### # ### ## # # # ####### ###### # ############ #### ####### # # ###### #### ############ # ############### # ###### ########################### # ########## ### ##### # ##################### ##### ## ### ## ## ##################### #### ## ### # ################################ ############### # ####### ##### # # ############# #### ############## # ##### ##### ######### # ##### ######## # ########################################## # ## #### # # # ## # # ##### # # # ############################################################################## ####### # # ## ############################# #### # ### ############## # ################## ################################# ############ # # ## ############################################################# ########### ### # # ##### # # ########## ############################## ####### # ### ##### # ################################### ####### # ###### ############### # ######## ################################ ##### #### # ############################# ## # ########################################################################################## # #### ##### ######################################################################################################################################## # ##### ##################### # #### ### ### # ##### ####################### ###### ################################### ###################### ###### ######### #################################################### #### ################################ # ### # ########## ########################################### ## ## ####### # # ## # # # ########## ############### ######################################### ##### #### #### # #################### # ######## # ## # ######## # #### ###### # ##### # #### # ## ##### ######### # # ## # ####### # # # ################################## # ################################################################################################### ## ### ## ## ################################## ### # ################### ## ########## ################# ## # ################### # # ### ## # ########## ############################### # # # ### # ############## ######### ## #### # ####### # #################################################### ### ### ###### # ### ############################# ######### #### # # ####################### ### ## # ####################### ##### # # ############################################################################ # ### # ####################### # ## # ########################################### ################## ##### ################################################################################################################################################################ #### # #################################################### ### # ################################# ###### # ### ### ############################################################################### ################### ## # # # # ## ##################################### ####### ## ####### ####################### ########## # ##### # ######################################################################## # # #### # ################################################################################################### # ##### ########### # ################## ######## # # ###### ### ########################################### ###### ####################################################################################################################### # # # ### ########################################## ## ## # ###################################### ###### # # ########################## ####################################################################################################### ### ## ### # # # # ### ###################################################################### ##### # ## # # ############################ ###### # ##################### ################ ###### # ##### ########################################################################################################### ## ################## # #### # # ############ ###################################################################################################################################################### # ## #### # # # # ## ############################################################################## ######### # ####### ####### ### ##### ##### ############################################################################################################################ ######### ###### ################## # ########## # ### ## ######### # ##### ## ##################################################################################################################################################################################################################################################################### # # # # # #### #### ## # ## # # ############################## ######## ############### ######## # # ################ # ## ################# ### ##### ### ######################################################################################################################################## ####### ######### ############## # ################################################ ############### ## ################################################################### ######## ################################################################################################################ ## ### ##### ############ # ############### ###### ### ################### ################################################ ########################## ### # # ########### ### # ###################### ######################################################################################## #################################################### ###################################################### # ######################################################################################################################### ##################################### ### ############################################################################# ################################################# ################################### ## # # ############################################## ### ####################################### # # ######################################## ############### ################################################################################################################################### # ### ## ############################################################## # ######## # # ################# # ####### #### ############ ############################################################################################################################################## # # ######### ### ###################################################################################################################### ################################################################################################################################################################ ######### ########################################################################################################################################################### ####### # ###################################### ############################################################### ### ####### ########################################################### ## ##### ############ ################################## ######## ########### ######## ##### ### # ########## ##################### ################################## ################################ b) Regional divisions and grid blocks ################################################################# ######################################################################################################## ###### ################################################ ########## ########### ## ### ##### ######################## ##################################################### used in the analysis. ########################################################################################################################################################## ######## ########################################################## ##################### ###################################################################################################################################### ################################################# Central Australia ## Central & Southeast Qld North Queensland Northwest Australia Central & Southwest WA South Australia Southeast Australia Tasmania

Figure 3.1 a) Distribution of H. leucogaster sightings across all Atlas Periods within the Australian Bird Atlas Data, including 2000 from the Historical Atlas (1901 – 1976), 5480 records from Atlas 1 (1977 – 1981) and 6933 records from Atlas 2 (1998 – 2001). b) Regions used in the analysis and grid blocks in which there were ≥ 30 sheets for all three periods.

68 Chapter 3 Distributional Variation

Using these data, I calculated a measure of the relative frequency of H. leucogaster within each 1° block, the Occupancy Index (OI):

No. of record sheets per grid block in which H. leucogaster and any of the other 21 aquatic species were present

Occupancy Index (OI %) = x 100

No. of record sheets per grid block in which any of the 21 aquatic species were present (with or without H. leucogaster)

Exploratory analyses had revealed that there were three types of bias within the raw data:

1. Increased sampling effort within woodland and open habitats within Atlas 2

(see alsoGarnett et al., 2002), associated with an a priori objective to review

the status of woodland bird species (Blakers et al., 1984; Clarke et al., 1999).

This was controlled for by filtering records using the aquatic species within

the OI as described above.

2. Repeat sampling of an area within the same overall sample period (where a

sample period could be anywhere between one day or many years), potentially

leading to inadvertent re-sampling of individuals, and an artificially inflated

sheet count when considering data over long periods, rather than when

measuring seasonal fluctuations in OI (Blakers et al., 1984). Since this

recording redundancy would occur with equal frequency in both H.

leucogaster and the other aquatic species, we considered that it would not

affect the OI.

69 Chapter 3 Distributional Variation

3. Over-sampling of areas close to human settlement (Clarke et al., 1999), which

was controlled for by expressing the occupancy index as a proportional rate.

3.2.3 Data Analyses

The OI was calculated for each grid block in each Atlas Period. Two types of comparison were used to assess changes in the OI over time. First, the mean OI values were compared among the three Atlas Periods in two-factor (time x space)

ANOVA’s, where the temporal treatments were the three Atlas Periods, and the spatial treatments were either the eight regions shown in Fig. 3.1b, or a two-part division into blocks that were coastal (if they intersected the geographic coastline) or inland. Of the 317 blocks used in the analysis, 123 were coastal and 194 were inland.

Pairwise differences among mean values were tested with LSD comparisons using the

“GLM” and “lsmeans” procedures in SAS (1996).

Second, the magnitude of change in occupancy within each block between pairs of

Atlas Periods was calculated based on positive and negative density differences.

These were: Time Period 1 - the difference in OI between the Historical Atlas and

Atlas 1 (i.e. between 1901–1976 and 1977-1981); and, Time Period 2 - the difference in OI between Atlas 1 and Atlas 2 (i.e. between 1977-1981 and 1998-2001).

Two-factor (time x space) ANOVA’s were used to test whether the change in OI per block was significantly different between the two Time Periods, and: 1) between the eight regions, 2) between coastal and inland blocks. Single-factor ANOVA was used to compare the change in mean OI between “urbanised” (N=13) and “non-urbanised”

(N=110) coastal blocks between the Historical Atlas and Atlas 2. Blocks considered

70 Chapter 3 Distributional Variation

‘urbanised’ were identified using GIS, and were those in which a significant proportion of the block contained urban activity or centres of high population density as classified by the Australian Surveying and Land Information Group (AUSLIG,

2001) (Fig. 3.2).

Urbanised Non-Urbanised

Figure 3.2 Location of ‘urbanised’ (N=13) relative to ‘non-urbanised’ (N=110) coastal blocks.

Density classes were also established within each Atlas Period by allocating OI values to classes based on the quartile range of the data across the Atlas Periods: 0%

= Absent, <3% = Low (1st quartile), 3 - 17% = medium (2nd and 3rd quartile), ≥ 18%

= High (4th quartile) (Fig. 3.3). For graphical purposes, a block was considered to

71 Chapter 3 Distributional Variation

show a “decrease” in occupancy between two Atlas Periods if the density class

changed from a higher to a lower category across the Time Period, and an “increase”

if the shift was from lower to higher.

400 le e rti til a tile ar u ar 350 u rd Q u t Q Q s 3 th & 4 1 nd 300 2

250 y nc

ue 200 q e r F 150

100

3 17 30 40 73 ent s b A Occupancy Index within blocks across Atlas Periods

Figure 3.3 Frequency distribution of Occupancy Index values (%) per block across all Atlas Periods.

72 Chapter 3 Distributional Variation

3.3 Results

At the continental scale, H. leucogaster was present in 57.4%, 62.2% and 62.8% of the 317 blocks within the Historical, Atlas1 and Atlas 2 periods respectively. In all

Atlas Periods, high OI blocks were concentrated along the coast, and Tasmania was the only region in which all blocks showed some level of occupancy during each

Atlas Period (Fig. 3.4).

Most of the variation in OI occurred between regions rather than between Atlas

Periods (Fig. 3.4 & 3.5, Table 3.1a), although the pattern of occupancy within regions differed between Atlas Periods (Fig. 3.4). In most cases mean OI levels were most similar between the Historical Atlas and Atlas 2. In five out of eight regions, mean OI was higher during Atlas 1 than the other Atlas Periods (Fig. 3.5a), particularly within coastal blocks (Fig. 3.5b), which showed large increases in occupancy, while inland blocks showed smaller but marked declines. With the exception of Central and

Southeast Queensland and Tasmania, all regions displayed a small (1-4%) net decrease in OI between the Historical Atlas and Atlas 2 when considering all blocks

(Fig. 3.5a), although there was no statistically significant difference in OI across Atlas

Periods (Table 3.1a).

73 Chapter 3 Distributional Variation

Occupancy Index ()%

Absent Historical Atlas 0.1 - 2.99 (1901 - 1976) 3 - 17.99 ≥ 18

Atlas 1 (1977 - 1981)

Atlas 2 (1998 - 2001)

Figure 3.4 Occupancy Index per 1° block for each Atlas Period. Pie charts quantify the proportion of blocks within each category.

74 Chapter 3 Distributional Variation

A significant interaction effect between Atlas Period and Location (Table 3.1a) showed that the pattern of change across time differed between coastal and inland blocks: coastal blocks showed elevated OI values during Atlas 1 (Fig 3.5d; one-way

ANOVA F=3.26, d.f.=2, 366, P=0.039), whereas inland blocks had greatest OI values during the Historical Atlas (one-way ANOVA F=2.65, d.f.=2, 579, P=0.072).

Overall, there was a general trend of decline in OI over time within inland blocks, especially within northern regions (Fig. 3.5c). In these analyses, blocks containing OI values greater than 18% were aggregated. Treating them as a single entity obscures the fact that there was a net reduction in the proportion of high OI blocks between the

Historical Atlas and Atlas 2. This reduction was statistically significant (χ2 = 7.44, df=1, P =0.0064) among inland blocks (Historical Atlas = 7.22%, Atlas 2 = 1.55%), but not among coastal blocks (χ2 = 1.202, df=1, P >0.05; and Table 3.2). I hypothesized that the reduction in coastal blocks may be due to the influence of increased urbanisation (Fig. 3.2). Although the mean decline in high OI blocks between the Historical Atlas and Atlas 2 was several orders of magnitude larger in urbanised (mean = -3.10%, s.e. ± 1.30%, median = -1.44%) than non-urbanised

(mean = -0.61%, s.e. ± 1.19%, median = -0.52%) blocks, this difference was not statistically significant (t-test = -1.413, d.f.= 37.9, P=0.166).

75 Chapter 3 Distributional Variation

a) All Blocks Historical Atlas (1901 – 1976)

25 Atlas 1 (1977 – 1981) c Atlas 2 (1998 – 2001)

cc 20 b) Coastal Blocks 35 c a 15 30 b ac d 25 c c 10 de e 20 5 b a 15 b

0 10

5 .

. .

a a a d a d t a a

i i i l i l i i

A A

l l l lia l lia s l l .

-5 .

a a a a a a u

and and

r r r Q r Q r r

ra ra ani l l

t t t t t t t t t t W W

s s s s s s s s s s t t 0 m t )

ns ns

u u u u u u u u s s s .

. .

a a d a d t a a a

e e

i i i A A A A l l a i i

A A %

l l l l l

s A A . . T l A ( l A

t t

a a uee a uee

h h h h u and and

r r ra r ra Q Q a a l l

t t t t ani heas

s s

t t t t t t t r r x

Q Q

u u u u W W e e t t

s s s s s s s t t m

t ns ns

o o o e o

n n & S.Ea & S.Ea

u u u u u w w h h s s out l l s t t

e e

S S S d S

h h e e a a r r

A A A a

S & S.W & S.W

A A r r l l C n C o o rt rt T t t t t uee

uee

h h a h a

t t s t s N N heas n n r r Q Q t t

u u e u e e e No No

n n

o o o c & S.Ea & S.Ea w w C C h h out l l e e

t t

S S n S h h r r a a S & S.W & S.W C C r r l l o o rt rt a t t a a N N n n p r r t t No No e e n c) Inland Blocks n C C e e ccu C C

O 16 c 14 c d) Coastal v Inland Blocks 12 20

10 16 8 b 12 6 b 4 8

a 2 aa 4

0 0 l d a n t . . -2 . a

d d

a a a a t l a a

l l

i i i i i i

A A

l l l lia l lia l l s nd nd . . In oas

Q Q

a a a a a a

u a a

r r r r r r

l l ra ra

t t t t t t t t t t

W W

s s s s A s s s s t t s s ns ns

-4 t

u u u u s s u u

e e

A A A A

A A l Au l Au t t uee

uee

h h h h

a a

t t t t s s heas S.Ea S.Ea r r

Q Q

u u u u

e e t t

o o o o n n & & w w h h l l out t t

e e

S S S S h h a a r r & S.W & S.W S r r l l C C o o rt rt t t a a N N n n r r t t e e No No n n C C e e C C

Figure 3.5 Occupancy Index (mean ± s.e.) within regions for each Atlas Period, and for coastal and inland locations. Letters show results of single factor ANOVAs (based on region) where the interaction terms (Time x Region) are not significant (see Table 1). Regions with the same letter are not significantly different (P>0.05).

a) Occupancy Index compared between the 3 Atlas Periods

Factors ANOVA P values

N blocks in Region or Level data set 1st Factor df 2nd Factor df Atlas Period Location Interaction All Blocks (Continental) 951 Atlas Period 2 Region 7 0.1038 0.0001 0.8806 Inland 579 Atlas Period 2 Region 6 0.0128 0.0001 0.1283 Coastal 369 Atlas Period 2 Region 7 0.0114 0.0001 0.6506 All Blocks (Continental) 951 Atlas Period 2 Location 1 0.0362 0.0001 0.0051

b) Change in Occupancy Index compared between the 2 Time Periods Region or 1st Factor df 2nd Factor df Time Period Location Interaction

All Blocks (Continental) 634 Time 1 Region 7 0.0002 0.6747 0.0059

Inland 386 Time 1 Region 6 0.0221 0.0121 0.6950

Coastal 246 Time 1 Region 7 0.0001 0.7889 0.0007

All Blocks (Continental) 634 Time 1 Location 1 0.0001 0.8354 0.0001

77 Chapter 3 Distributional Variation

Across the whole continent, the proportion of stable blocks relative to those increasing or decreasing was similar between Time Periods: between the Historical

Atlas and Atlas 1 the mean change in OI across all 317 blocks was 0.5%, between

Atlas 1 and Atlas 2 it was –1.6%, and between the Historical Atlas and Atlas 2 it was

–1.1% (see also Fig. 3.6). However, there were considerable shifts in the geographic pattern of occupancy. Coastal blocks showed much more change than inland blocks

(Fig. 3.7, Table 3.1b): between 1901– 1976 and 1977-1981 H. leucogaster increased in frequency in coastal blocks while decreasing in inland blocks, whereas the reverse occurred between 1977-1981 and 1998-2001 (Fig. 3.7d). Within coastal blocks there was also considerable variation among regions. Between 1977-1981 and 1998-2001, the Southeast region showed a small increase in OI within inland blocks. However,

Central and Southeast Queensland was the only region where there was a marked increase in OI within inland blocks (Fig. 3.7c). Notably, H. leucogaster density, which has been considered stable in the remote north and tropical regions (Dennis and

Lashmar, 1996), here shows a sustained decline in Northwest Australia and North

Queensland (Fig. 3.5 & 3.7) as well as localised extinction within some inland blocks

(Fig. 3.8). Central Australia was the only region in which H. leucogaster apparently became locally extinct (Fig. 3.8).

78 Chapter 3 Distributional Variation

Category of Change in The Historical Atlas to Atlas 1 (1901-1976 to 1977-1981) Occupancy Index Between Time Periods All Blocks Sta ble Increase Decline

Coastal Bl ock s

Inland Bl ock s

Atlas 1 to Atlas 2 (1977-1981 to 1998-2001)

All Blocks

Coastal Blocks

Inland Blocks

79 Chapter 3 Distributional Variation

The Historical Atlas to Atlas 1 a) All Blocks (1901-1976 to 1977-1981) 10 Atlas 1 to Atlas 2 (1977-1981 to 1998-2001) 8 b) Coastal Blocks 6 4 20

2 15 0 10 -2 5 -4 0

-6 )

) %

% -8 -5

( (

x x

e e . . .

a d d d a

a a a a a d -10 d t a

i i ii

i i i l ll i i -10

l A lia lia l ll A

l l l

d l d n n n

. .

a a

a a a

Q a a a

n n u

r r l ll rr

r r r r ani

Q a

I I

t t t t tt

t t t t tt t

s A

s s s s

s s s s s s m

t m

y y

ns n

t t

u u u u

u u u a u s s s s -15

e s

c c

a A e a A A e

A A A E A

e a

n n A

l T l A l T

t tt

uee u e

h h h

a a a S

s s

t t t t heas h

. .

r r

Q Q

e e

u u u

p u p t t t

.. .

S S -20

& u

d a

a a a d a t a

o o o

n o

u u h w

& S.Ea l h w l ii

i i i i i

o l ll o ll

l l l t tt l

n s

e ..

S S S

h S

r &

rr h & n

c c

a a Q a

a a a a

S t u tt a

r r r C r o r C r l & S. ll a r o ll

r rr

c c

tt tt

t t t t tt W t

a s a

o o

N N

s s

s s s n n tt s

r m

O O n t

t tt

N a N a u

u u u

u e e s

n n

E A a

A A A

n A n C C

e ..

i i

h h h W h

C S C

t t t .. t h

e e

u u u u

g g u

o o o o

n n

S S S h S

a rr

tt S

a a

rr ll

c) Inland Blocks o rr

h h a

e N

C C

6 n

4 a d) All Blocks by Location a 2 6 a a a 0 4

2 -2 0 -4 -2 -6

-4 s bb k

.. .

a a a a d d a B

a t

i i i ii ll i

ii s

l l l

-8 ll l

n ll -6

k d

a a a a Q a

a a

r r r r ll r

c n

r cks

t t t tt tt t tt W

s A

o o

s s s s l s s l

s tt l

u u u u a u

n u s

e s I

A A A A A l A E l

e a

ll a

u tt e

h h h h t t

t t t t

s .. h

s s

Q t

u u u u

tt e

a S

o o o o

w o

S S S & S

a h

r ll

C o

N o

e N

C Figure 3.7 The change in Occupancy Index (mean ± s.e.) between Time Periods, within regions and for coastal and inland locations. Letters show the result of single factor ANOVAs for regional differences in the magnitude of change where the interaction terms (Time x Region) are not significant (see Table 3.1). Regions with the same letter are not significantly different (P>0.05).

80 Chapter 3 Distributional Variation

Newly Occupied

Locally extinct Dams or impoundments

# # ### ##

# # ## # # #### # # # # # # # # ## # ### # # ## # # ### # ## # # # ## ### # # ## # # # # # ## # # # ## # ## # #### # ## ## #### ###### # ## # # ## ## ## # ###### # ## ### ##### ## ################ ############# # ##### # # # ### #### # # # ####### ## ### ## # # # ## # # # # # # # ### ## # # # # ### ## # # # # # ## #### # # ### ## # ### # # # # ##### ## ##### # # # # # ##### # ##### ############### # ### #### #################### # ## ### ### # ####### # ### ######## # ### # ############### # ###### ## ##### # # ### ## # ##### # ######## ##### ###### ##### # ######## # # ## ## ######## ### ## # ## ## ### # # # ### # ### ##### # #### # ## ######## ########## ##### ######### ### ## ######## ############ ### # #################### #################### # ### # # #### ######### # # ###### # # ############### ################# ## ############ # ############ # ####### # # #### ## ##############

Table 3.2 The percentage of coastal blocks (N=123) within each Atlas Period with high Occupancy Index values (≥18%) subdivided into three categories.

Total <18 % 18-25 % 26-40 % 41 % ≥ ≥18 %

Historical Atlas 65 18.7 11.4 4.9 35

Atlas 1 64 9.8 16.3 9.8 36

Atlas 2 71 13.8 12.2 2.4 29

81 Chapter 3 Distributional Variation

3.4 Discussion

3.4.1 Detecting ‘macro-scale’ changes in frequency and distribution

As would be expected on the basis of H. leucogaster biology, areas of highest OI were associated with the coastal zone, or those blocks that included large bodies of perennial water, and therefore suitable nesting and feeding sites (Newton, 1991;

Raphael et al., 2002). Additionally, it was expected that there would be differences between regions in OI, as well as in the number of blocks occupied, reflecting differences in the suitability of the local environment across the continent (Brown et al., 1995). Habitat suitability will depend on both potential food sources and potential nest sites (Newton, 1991). No distinction was made in the raw data with respect to breeding and non-breeding individuals, however Newton (1991) suggested for raptor species that even where nest sites are not available, areas will support non-breeding birds if their are sufficient food resources,.

Raptors especially when a population is large, have a ‘floating sub - population’ comprising non-territorial, non-breeding birds of both sexes which live unobtrusively both within and outside of existing territories (Olsen, 1995; Hunt, 1998). Sedentary adult pairs maintain a small territorial range during the breeding season, however the home range over which H. leucogaster may forage is estimated to be large (up to

100km2) and proportional to the available food supply (Marchant and Higgins, 1993).

Long-range movements of floaters may have accounted for the shifts in geographic distribution apparent in this study (see later discussions).

82 Chapter 3 Distributional Variation

In spite of these shifts, there appeared to be overall stability at the whole-continent scale, where the proportions of increasing, decreasing and stable blocks remained similar over the complete timeframe of the analysis (Fig. 3.6). Spatial stability over

20 year periods was reported for conspecific comparisons in the American Breeding

Bird Survey (Brown et al., 1995), and for a large number of Australian species over the 20 year period between the first and second Australian Bird Atlases (Brown et al.,

1995; Garnett, 2001). However, in the present study there were considerable changes in the frequency of H. leucogaster at the regional level and when coastal and inland areas were compared. This highlights the importance of distinguishing between processes at different geographical and temporal scales (Flather and Sauer, 1996;

James et al., 1996; Kirk and Hyslop, 1998). Significant regional variation for the change in reporting rate of H. leucogaster between Atlas 1 and Atlas 2 was also reported by Garnett et al. (2002a) and Barrett et al. (2003). However, contrary to this analysis, they also found an overall increase in the reporting rate of H. leucogaster within Australia between these Atlas Periods, and within specific bioregions. Their method of analysis used standardisations for survey effort that had been designed for more general purposes. The method described here (using records in which a range of other aquatic species were present) is more targeted, as it is not biased by an emphasis on passerines in the later atlas, and hence probably more sensitive to changes in H. leucogaster abundance within broadly suitable habitat.

Effectively I compared data collected over an 80 year period in the Historical Atlas with data collected over a 5 and 4 year period in the Atlas 1 and Atlas 2 periods respectively. The use of Historical data in such comparisons has been criticised on the

83 Chapter 3 Distributional Variation

basis that there are likely to be fewer observations, and that these may be skewed toward actively collected museum specimens or logistically accessible regions

(Franklin, 1999). There were many fewer H. leucogaster observations in the historical data than in either Atlas 1 or Atlas 2. However, the choice of 1° blocks with at least

30 record sheets for analysis, together with the OI that provided a correction for aquatic survey effort, constitutes an approach that is more conservative than many other studies based on Atlas data (see for example: Robertson et al., 1995; James et al., 1996; Reid, 2000; Garnett et al., 2002; Barrett et al., in press), and appears to have compensated for any such bias, as there is no significant difference in occupancy between Atlas Periods that could have been attributed to differential survey effort

(Fig. 3.5 & Table 3.1a).

3.4.2 The effect of climatic variability.

The ‘boom-bust cycles’ (Kingsford et al., 1999) associated with climatic influences of the El Niño/Southern Oscillation cycle (ENSO) have been implicated in affecting many aspects of species’ biology (e.g. Jaksic and Lazo, 1999; Brichetti et al., 2000;

Jaksic, 2001; Roshier et al., 2001; Zavalaga et al., 2002), and within Australia are strongly linked to the filling and productivity of ephemeral wetlands, and therefore the availability and abundance of wetland and terrestrial prey species (Ridpath and

Brooker, 1986; Robertson, 1987; Kingsford, 2000; Roshier et al., 2001).

A strong El Niño event during 1977/1978 affected the majority of Australia (Flood and Peacock, 1997). Between 1979 and 1981 large sections of eastern Australia and southwest Western Australia recorded extremely low rainfall (in the 0-10th

84 Chapter 3 Distributional Variation

percentile) (Flood and Peacock, 1997). The Atlas 1 period (1977-1981) coincided with consistently negative Southern Oscillation Index (SOI) values, while in contrast the SOI was consistently positive during almost all of the Atlas 2 period (1998-2001), suggesting above average rainfall in many areas throughout this time (Department of

Natural Resources & Mines, 2003). This pattern agrees with findings of rainfall analysis as part of the National Land and Water Resources Audit (Garnett et al.

2002b).

Accordingly, widespread drought conditions associated with the Atlas 1 period are consistent with a movement of H. leucogaster to coastal regions (Fig. 3.4 & 3.7) and may explain the very high OI blocks in the north and eastern part of the country at this time (Table 3.2 & Fig. 3.4), as well as the expansion of birds into coastal blocks in South Australia, Central and Southeast Queensland, Northwest Australia and

Central and Southwest Western Australia, where food supplies or breeding territories may have been more reliable (Fig. 3.4).

Coastal block increases may therefore be the result of an influx of non-breeding birds in response to drought conditions rather than an increase in the stable breeding population (Olsen, 1995; Gaston et al., 2000), with floating birds returning to previously occupied inland areas when conditions improved. This cyclic movement is consistent with the change in occupancy between coastal and inland blocks seen when comparing Time Periods (Fig. 3.7d).

85 Chapter 3 Distributional Variation

3.4.3 Human impacts and long-term trends

Haliaeetus leucogaster is known to populate almost all artificial water impoundments in Australia, some of which are stocked to support recreational fishing (Marchant and

Higgins, 1993; Olsen, 1995). The regulation of river flows through impoundment and diversion could have influenced patterns of occupancy within inland blocks (Fig. 3.8).

The immediate effect of regulation is to remove the variable flooding pattern of transient wetlands to create permanent waters (Kingsford, 2000; Roshier et al., 2001), which offsets the effects of drought on prey availability and breeding success in several raptor species (Robertson, 1987; Brown, 1996; Bryan et al., 1996).

Additionally, it has been suggested that permanent water may favour some raptors as there can be a greater concentration of prey around these areas in response to dry conditions (Olsen, 1995).

Density increases in inland Northwest Australia and central Western Australia, as well as ‘newly occupied’ blocks, show some relationship to the presence of dams

(Fig. 3.8), although in many cases a dam was in place long before colonisation by H. leucogaster (Geoscience Australia, 1990). To account for sightings of itinerant birds, blocks were considered ‘newly occupied’ where the OI had increased from zero during the Historical Atlas to greater than 3% in Atlas 2.

As important food species, colonisation of waterways by Carp Cyprinus carpio

(Koehn et al., 2000), and the spread of the European Rabbit Oryctolagus cuniculus have been indicated as assisting the increase of inland populations of H. leucogaster

(Olsen, 1995; 1998). On the other hand, within Victoria and South Australia, land use

86 Chapter 3 Distributional Variation

change, including clearing for agriculture, grazing, forestry or other industries, as well as direct disturbance, has been blamed for a loss of breeding sites and decreased breeding success in H. leucogaster (Clunie, 1994; Dennis and Lashmar, 1996).

Irrespective of this, the trend for urbanised coastal blocks to show a reduction in OI relative to non-urbanised coastal blocks was not statistically significant.

Within Tasmania, there were large declines in the south, in spite of the overall trend of increase for the whole region (Fig. 3.8), because most northern blocks showed an average increase in OI. Without further research it is not possible to show if increases in the north might be due to an influx of birds from the south, or to what extent the breeding population as a whole might have changed. Various combinations of habitat clearance, human disturbance, accidents and direct persecution have been suggested as directly affecting distribution in recent years (Mooney and Brothers, 1986).

Similarly, there are no clear explanations for the pattern of decline in inland Northern

Australia and North Queensland.

Effectively this analysis represents measurements of change in occupancy over a 100 year period, or only about 20 generations, based on approximately 5 years at first reproduction (Marchant and Higgins, 1993). The impression of occupancy is dependent on both the temporal and physical scale of analysis. For example, the cyclic effect of drought-induced disturbance on H. leucogaster in Australia would be very difficult to detect without moderate to large temporal scale data sets, as over the complete timeframe (1901 – 2001) the flux in drought-induced occupancy is virtually indiscernible. The large shifts in abundance, both within individual blocks and at the

87 Chapter 3 Distributional Variation

regional level between Atlas Periods, may have been perceived as local extinction within a number of Australian states, irrespective of a relatively unchanged range extent at the continental scale. This is not to suggest that localised colonisations or extinctions resulting from water impoundment or urbanisation are not locally significant. However, they appear to be having little impact at even the regional scale.

At the broadest spatial scale, and when considering the complete timeframe, most shifts in OI or geographic extent appear driven by the effect of climate. Should this study have been conducted within the span of a single Atlas Period a different impression of abundance would have been obtained.

88 Chapter 4 Discrimination of Sex

Chapter 4 - Discrimination of sex using genetic and

morphometric techniques

4.1 Introduction

To facilitate the morphometric analysis it was necessary to reliably sex Haliaeetus leucogaster individuals. H. leucogaster is monomorphic for plumage colouration in adult birds, however like many raptor species, it displays reversed sexual size dimorphism, with female birds larger than males for a number of morphometric characters (Baker-Gabb, 1984; Marchant and Higgins, 1993; Ferguson-Lees and

Christie, 2001). Therefore, it is possible that birds could be sexed based on morphometric analyses if the relationship between sex and body size were quantified.

This would require a sufficient number of independently sexed individuals, which has previously been hard to find, excepting a small number of museum specimens sexed by dissection or laparoscopy. For other individuals, sex has often not been known, or has been allocated as a ‘best guess’. The advent of genetic sexing methods suitable to a wide range of bird species (Griffiths et al., 1998) has provided the opportunity to definitively assign sex in living birds as well as preserved specimens.

A complication to the application of morphometric analysis may arise if body size varied geographically. Such variation is known to occur in many bird species (Zink and Remsen, 1986). For example, a large number of species vary in size latitudinally in response to ecological or environmental effects (Zink and Remsen, 1986), with

89 Chapter 4 Discrimination of Sex

clines of character variation often maintained through differing regional selection pressures (Fitzpatrick and Dunk, 1999).

This chapter had a number of aims: first to determine if H. leucogaster could be sexed genetically using existing techniques, second, to use this information to develop a discriminant function using morphometric measurements to sex birds within the

Australo-Papuan part of their range, and third, to determine if this model is robust to the effect of latitude.

4.2 Methods

4.2.1 Data collection

Samples (N=37) were obtained from a range of sources (see below). Thirty-two birds were adult, two were juvenile, one was immature and two had not been aged. Fifteen morphometric measurements were collected from most birds (Table 4.1).

Measurements were taken on the right side of the body using vernier calipers or metal ruler, by a number of field workers following the descriptions of Olendorff (1972),

Bortolotti (1984) and Masterov (2000). The full set of measurements could be taken in less than 10 minutes under field conditions.

90 Chapter 4 Discrimination of Sex

Table 4.1 Description of morphometric variables measured (those asterisked were not included in the analyses due to suspected bias or high levels of missing data).

Variable Number of birds % of data from which from measurements replacement were obtained values

• Weight (g) * 33 10.8 • Wing cord (mm) 35 5.4 top of the wrist to the tip of the longest primary • Head length (mm) 34 8.1 back of the head to the tip of the bill • Tail length (mm) * 35 5.4 • Culmen tip/cere (mm) 33 10.8 tip of the bill to the front edge of the cere • Culmen tip/base (mm) 34 8.1 tip of the bill to the plumage edge at the front of the head • Culmen width (mm) * 13 35.0 width at the cere

• Culmen depth (mm) * 13 35.0 depth at the cere • Tarsal joint (mm) 35 5.4 • Tarsal length (mm) * 37 0.0 • Tarsal breadth (mm) 35 5.4 taken at the midpoint • Tarsal width (mm) 36 2.7 taken at the midpoint • Halux claw length (HxL, mm) 36 2.7 • Halux claw breadth (mm) 36 2.7 depth of the claw at the intersection with the toe • Halux claw width (mm) 35 5.4 width of the claw at the intersection with the toe

91 Chapter 4 Discrimination of Sex

Genetic material (blood, tissue or feather) was obtained from whole individuals (18 birds) or from museum skins/mounts (19 birds). Three of the former were live birds held by fauna parks and six were frozen carcasses held by museum or government agency tissue banks (e.g. CSIRO). Nine birds were caught in the wild using a noosed- fish technique (see Appendix 1) as part of the sample for the population genetic analysis (Chapter 2).

4.2.2 Genetic sexing

DNA was extracted following the protocols described in Chapter 2.2.3.2, and PCR amplicons were prepared using the P2 and P3 primers (Griffiths and Tiwari, 1995) which simultaneously amplify the CHD-W (chromobox-helicase-DNA-binding) and the CHD-Z genes (Griffiths et al., 1998). PCR products were cut using a HaeIII restriction enzyme which cleaves the Z band. Female birds show heterogamety (ZW) whereas male birds do not (ZZ) (Norris-Caneda and Elliott, 1998). Therefore, females have one strong band, resulting from the uncut W fragment and two shorter bands from the cut Z fragment. Males have two smaller bands. Laparoscopic data were available for a number of museum specimens (six females and six males) which were used to validate the genetic sexing protocol.

PCR protocols followed those of Norris-Caneda & Elliott (1998) with the following modifications: PCR reactions included: 3.1µl 10x polymerase reaction buffer (Fisher

Biotech), 2µl 25mM MgCl2 (Fisher Biotech), 0.1µl Taq DNA Polymerase (5 units/µl

– Fisher Biotech), 10mM each of dATP, dGTP, dCTP, dTTP, (Fisher Biotech), 1µl

(10µM) of primer P2 (5' TCTGCATCGCTAAATCCTTT 3'), 1µl (10µM) of primer

92 Chapter 4 Discrimination of Sex

P3 (5' AGATATTCCGGATCTGATAGTGA 3'), and between 30-100ng of DNA template adjusted to a final volume of 25µl with ddH2O. All PCR runs contained a negative control. Temperature cycling was as follows: denaturation at 94°C for 1.5 minutes, then 45 cycles of 94°C denaturing for 30 seconds, 53.7°C annealing for 30 seconds, 72°C extension for 30 seconds followed by a final extension of 5 minutes at

72°C. 24µl of each PCR reaction were cut with 2 units of the restriction enzyme

HaeIII (New England Biolabs). Digests were performed for 1 hour at 37° C following the manufacturer’s recommendations. Digested product was visualised and compared to a known marker on a 3% agarose gel containing ethidium bromide run at 60 V for

90 minutes in 0.5X TBE buffer.

4.2.3 Data analysis

Prior to multivariate analyses, missing data were replaced with an average derived from the total sample. Each variable and group (male and female) was checked for deviation from a normal distribution using normal probability plots and measures of skewness and kurtosis (Tabachnick and Fidell, 2001). Width at cere and depth at cere were dropped from the analysis due to insufficient data. Both tail and tarsal length contained a high proportion of data at the extremes of their distributions or had outliers. Data transformation (Quinn and Keough, 2002) did not improve their distribution, and data were analysed with these variables removed. Weight was not included in the analysis as it was considered unreliable due to the inclusion of wild and captive individuals, and may reflect condition or seasonal variation (Smith and

Wiemeyer 1992; Morrison and Maltbie 1999). Among the 10 morphometric

93 Chapter 4 Discrimination of Sex

characters, the percentage of missing values ranged from 2.7% to 10.8% across variables (Table 4.1), and 21% of individuals contained missing data.

Principal component analysis (PCA; SAS, 1996) based on a correlation matrix (Quinn and Keough, 2002) of the 10 remaining morphometric characters (Table 4.1) was used to determine if the data fell naturally into the genetically determined female

(N=18) and male (N=19) groups based on morphology, and also whether there was any geographic patterning to body size. The relationships between the first Principal

Component score and both latitude and longitude were tested for males and females separately using least-squares linear regression (SAS, 1996). The independence of latitude and sex effects on the first principal component score was tested using two- factor ANOVA (Sex x Latitude Group), using three latitude groups defined as: North

(-3°S to –18°59′S), Mid (-19°S to –38°59′S), and South (-39°S to –50°S). Latitude divisions were determined based on climate or potential geographic barriers. The effect of taking measurements from museum skins compared with other sample types was tested in males and females separately using single factor ANOVA. Museum skins comprised 61% of females and 42% of males. Pairwise differences among mean values were tested with LSD comparisons using the “GLM” and “lsmeans” procedures in SAS (1996).

A discriminant function to allocate sex was derived using the “DISCRIM” procedure, with the “CANONICAL” option, in SAS (1996). Data were tested for homogeneity of

“within” covariance matrices using the “POOL=TEST” option. A jackknife procedure was used to check the fit of the model using the “CROSSVALIDATE” option (SAS,

94 Chapter 4 Discrimination of Sex

1996). Discriminant function (Di) scores for each individual were calculated by multiplying the standardised value (zi) for each morphometric character by its standardised discriminant coefficient (di) then adding the products for all morphometric characters (Tabachnick and Fidell, 2001). Therefore, Di = zi1di1 + zi2di2

… + zi10di10; where, zi = (raw score - x )/ standard deviation. Individuals with Di > 0 were classified as female, individuals with Di < 0 were male.

4.2.4 Results

The genetic sexing technique was 100% accurate in birds that had previously been sexed by laparoscopy. An example of observed banding patterns is given in Fig. 4.1.

Comparisons of morphometric sample types revealed that in females mean tarsal breadth was smaller for museum skins than for other sample types (one-way

ANOVA; F=11.8, d.f.=1,17 P=0.003), and tarsal joint was larger (F=4.9, d.f.=1,17

95 Chapter 4 Discrimination of Sex

P=0.04). Tarsal width was relatively smaller in male museum skins (one-way

ANOVA; F=5.22, d.f.=1,18 P=0.04).

Principal Component Analysis of the morphometric data gave a clear split between males and females (Fig. 4.2a). Over 60% of the variation within the sample was explained by the first three principal components, with more than half of this due to the first principal component (Table 4.2). All variables had higher values for females than for males (Fig. 4.2b). Halux claw length (0.83), tarsal width (0.75) and wing chord (0.67) correlated most highly with the first principal component (Table 4.2).

The Discriminant Function model effectively discriminated between male and female groups (Pillai’s Trace; F=8.3, P<0.0001), with mean discriminant scores of –1.19 and

1.79 for males and females respectively. The strength of loadings associated with each variable generally showed a similar pattern to the PCA (Table 4.2). The linear function that best discriminated between male and female H. leucogaster was:

Di = 0.127(WC) + 0.158(HL) + 0.435(CTC) – 0.023(CTB) + 0.532(TJ) +

0.291(TW) – 0.100(TB) + 1.146(HxL) – 0.115(HW) + 0.209(HB) where an individual’s raw morphometric value for each character was standardised using the values given in Table 4.2 before inserting it into the equation. Linear functions based on fewer measurements were examined but produced a marked reduction in successful discrimination.

The Discriminant Function correctly identified 100% of 37 individuals. Re- classification following the jackknife procedure correctly identified 92% of

96 Chapter 4 Discrimination of Sex

individuals, mis-classifying one female and two males (Fig. 4.3). If individuals with discriminant function scores between –1 and 1 are considered to fall within a ‘zone of uncertainty’, as identified by the jackknife procedure, then 80% of 5 females and 60% of 5 males within this zone were classified correctly (Fig. 4.3). 100% of individuals outside this zone were correctly classified. In each case misclassified birds were from mid-latitude sites and none required replacement of missing data.

97 Chapter 4 Discrimination of Sex

Prin 2 Male Female

Prin 1

b) Prin 2 TB

HXL TW HB Prin 1 CTC TJ WC

HW CTB

Summary Statistics

Variable Abbr. Mean Std Dev Prin 1 Prin 2 DFA Wing cord WC 588.1 30.9 0.67 -0.20 0.28 Head length HL 111.9 5.4 0.59 0.49 0.34 Culmen tip/cere CTC 42.6 4.9 0.51 -0.09 0.27 Culmen tip/base CTB 51.9 4.2 0.65 -0.48 0.26 Tarsal joint TJ 23.3 2.2 0.54 -0.02 0.37 Tarsal width TW 16.2 2.2 0.75 0.02 0.47 Tarsal breadth TB 16.8 2.0 0.35 0.73 0.13 Halux claw length HxL 38.2 3.1 0.83 0.14 0.70 Halux claw width HW 9.5 0.9 0.33 -0.50 0.11 Halux claw breadth HB 7.2 0.9 0.65 0.00 0.34

Eigenvalue 3.67 1.32 % variation explained 37 13

99 Chapter 4 Discrimination of Sex

5 Female 4 Male 3 2 e 1 * Scor 0 Di * * -1 -2 -3 -4 -50 -40 -30 -20 -10 0 Latitude (degrees)

Figure 4.3 Plot of discriminant function scores (Di) against latitude for birds of known sex. Asterisked individuals were mis-classified in the jackknife procedure. Dotted lines define a ‘zone of uncertainty’ outside which an individual’s allocated sex could be reliably assigned using the discriminant function.

First principal component scores showed no significant correlation with longitude

(female: r2=0.13, P=0.142; male: r2=0.0002, P=0.96), and were significantly correlated with latitude for males (r2=0.37, P=0.006; and Fig.4.4) but not females

(r2=0.13, P=0.13). A two-factor ANOVA showed that the first principal component score was significantly related to both latitude group (F=4.07, d.f.=2,36 P=0.03) and sex (F=83.3, d.f.=1,36 P=0.0001). However, the interaction between sex and latitude was not statistically significant (F=0.17, d.f.=1,2 P=0.84).

100 Chapter 4 Discrimination of Sex

B Female 3 AB A Male 2

Prin 1 -1

-2

-3

-4 North Mid South

Figure 4.4 Principal component scores (Prin 1: mean ± s.e.) for male and female birds for each latitude group. Regions with the same letter are not significantly different (P>0.05).

4.3 Discussion

The genetic sexing tool using the P2 and P3 primer combination has been found to work well with a range of other raptor species (Norris-Caneda and Elliott, 1998). I also trialled the P2 and P8 primer combination (Griffiths and Tiwari, 1995) which has successfully sexed a small number of raptor species (Griffiths et al., 1998), but which failed in H. leucogaster. Although the P2/P3 is more time consuming than the P2/P8 primer combination, due to the restriction enzyme digestion step, it provided a 100% correspondence between the genetic sexing results and the results of sexing by either laparoscopy or dissection.

101 Chapter 4 Discrimination of Sex

With a predictive accuracy of between 92 and 100%, the derived discriminant function, based on 10 morphometric variables, appears to be a viable alternative to genetic sexing or laparoscopy (Garcelon et al., 1985; Morrison and Maltbie, 1999).

Futhermore, it yielded greater predictive accuracy than found in other studies of raptor species, which have reported misclassification rates of between 10 and 59%

(see for example: Coffey, 1987; Smith and Wiemeyer, 1992; Morrison and Maltbie,

1999; Balbontin et al., 2001).

Although body weight may effectively discriminate between male and female H. leucogaster (Marchant and Higgins, 1993), its exclusion from the discriminant function did not appear to reduce the model’s strength. Weight has often been used as a discriminator of sex (eg. Smith and Wiemeyer, 1992; Delehanty et al., 1995;

Balbontin et al., 2001), but its inclusion in predictive models has been criticised as it is difficult to standardise measurements to control for food consumption, seasonal fluctuation, or differences between captive and wild birds (Smith and Wiemeyer,

1992; Morrison and Maltbie, 1999). Variables that contributed most strongly to multivariate analyses based on morphometric measurements (halux claw length, tarsal width, wing cord, tarsal joint) (Table 4.2) are correlates of body size (Zink and

Remsen, 1986), particularly with respect to skeletal structure, and may therefore more reliably estimate size.

Differences in three tarsal measurements between museum skins and other sample types were probably due to shrinkage or tightening of tissue around the tarsal bone.

102 Chapter 4 Discrimination of Sex

Tarsal breadth had a very low loading in the discriminant function (Table 4.2), so it is unlikely to have affected group discrimination. Tarsal width and tarsal joint were both moderately loaded. Amongst males, 42% of samples were derived from museum skins. Smaller mean tarsal width in male skins ( x = 13.86 ± 0.47) compared to male non-skins ( x = 15.23 ± 0.37) could have slightly inflated the separation between sexes by decreasing the overall mean for males, and may have introduced some bias to the discriminant function. Similarly amongst females, 61% of measurements were derived from skins. The tarsal joint was significantly larger in female skins ( x =

25.35 ± 0.55) than female non-skins ( x = 23.37 ± 0.71), which may also have slightly increased separation between the sexes. However, this is unlikely to have had a large effect on the fit of the model.

Measurements from one immature and two juvenile birds were included in the derivation of the model. Given the Di scores for the three subadult individuals that were included in the analyses (0.80, 1.67, -2.88), it is unlikely that they introduced a bias. However, it is improbable that the model would be generally suitable for pre- fledged or nestling birds. Models developed for adults have been found inappropriate for sexing nestlings in at least one other raptor species (eg.Balbontin et al., 2001). An

‘adult’ was defined as a sexually mature individual. Age at first reproduction is approximately 5 years in this species (Marchant and Higgins, 1993).

According to the theoretical expectations of Bergmann’s Rule (Zink and Remsen,

1986), continuously distributed species become larger further from the equator. This pattern is attributed to thermoregulatory needs in colder climates, where a low surface

103 Chapter 4 Discrimination of Sex

area to volume ratio may be advantageous (Fitzpatrick and Dunk, 1999). Consistent with this, there appears to be a strong effect of latitude on body size in H. leucogaster.

Importantly, although northern birds were smaller than those from higher latitudes, this variation did not affect the ability of the model to discriminate between sexes, at least between -3°S and -50°S. This species’ natural range extends to 26°N. Fitting morphometric data to the model from a male and a female bird from the Malay

Peninsula (1°16′59″N) correctly classified the male and mis-classified the female.

Therefore, further testing and analyses of individuals within the northern part of the species range would be needed before morphometric sexing could be applied outside the Australo-Papuan region.

104 Chapter 5 Morphometric Variation

Chapter 5 – Evidence of a morphometric cline in the White-

bellied Sea-Eagle in the absence of mitochondrial DNA

structure

5.1 Introduction

Geographic structure in bird populations has been studied using a variety of methods including morphology (e.g. Fitzpatrick and Dunk, 1999), song (e.g.Tubaro and

Segura, 1995), plumage (Hughes et al., 2001) and molecular markers (Zink, 1997).

Interestingly, different methods do not necessarily reveal the same pattern of population structure. For example, the redpoll finch is genetically homogenous in

North America but shows extensive plumage and size variation geographically

(Seutin et al., 1995). Among studies with similar findings, this lack of concordance has been attributed to the recent evolution of morphological differentiation (Ball et al., 1988; Grapputo et al., 1998) and therefore the inability of genetic markers to reveal variation (Seutin et al., 1995; Storfer, 1996), selection for morphological traits in the presence of continued gene flow (Greenberg et al., 1998; Hughes et al., 2001), secondary contact following a period of isolation (Endler, 1977), or environmental induction (James, 1983).

Based on the mtDNA analysis, H. leucogaster has relatively low genetic diversity with little evidence for genetic structuring at least within the Australo-Papuan region

(Chapter 2). Only one region, Southeast Australia, differed from any other region.

105 Chapter 5 Morphometric Variation

This level of structure does not correspond with the apparent latitudinal body size variation revealed in both sexes (Chapter 4).

Therefore, the aim of this chapter was to review morphometric variation within the

Australo-Papuan region of the species distribution to determine if male and female birds differ between regions or according to latitude and longitude, and to compare morphometric variation with known patterns of genetic differentiation. This tested whether populations that appeared to be genetically differentiated also were differentiated morphologically. In addition, evidence of morphologically divergent regions may indicate limits to gene flow which were not detected in the mtDNA analysis.

5.2 Methods

5.2.1 Morphometric analysis

Morphometric data (N=95) were collected from fully-fledged birds. These data included the 37 birds previously sexed by the genetic methods described in Chapter 4.

Samples were obtained from a range of sources including the wild (28 birds), fauna parks (four birds), and museums or tissue banks (63 birds), and included fledged birds with adult (60 birds) and sub-adult (32 birds) plumage. Three birds had not been aged. Fifteen morphometric measurements were collected from most birds (Table

5.1). Measurements were taken on the right side of the body using vernier calipers or metal ruler, by a number of field workers following the descriptions of Olendorff

106 Chapter 5 Morphometric Variation

(1972), Bortolotti (1984) and Masterov (2000). Of the 95 birds, the 58 unsexed individuals were sexed morphometrically using a discriminant function model described in Chapter 4.

Two types of comparison were used to assess geographic variation in morphometric characters. First, individuals were assigned to the same latitude groups as used in

Chapter 4. Groups were defined as: North (-3°S to –18° 59′S), Mid (-19°S to –38°

59′S), and South (-39°S to –50°S). Latitude divisions were determined based on climate or potential geographic barriers. Two-factor ANOVA’s (Sex x Latitude

Group) on each of the 15 morphometric characters, and using all data (95 birds), were used to test whether the variation in morphology among male (N=50) and female

(N=45) birds was independent of the effect of latitude. To ensure that the inclusion of sub-adult birds or museum skins did not adversely affect variation, the effects of Age

(Adult vs Sub-adult) and Latitude Group (North, Mid and South) were tested for each character using two-factor ANOVA. Sub-adults were classified as fledged birds showing either juvenile or immature plumage patterns (Marchant and Higgins, 1993).

Mean differences between museum skins and all other sample types were tested using single factor ANOVA. In both cases, tests were performed for male and female birds separately.

107

Table 5.1 Morphometric variables and replacement values used for missing data in the calculation of the Principal Component Analysis, and morphometric variation compared between sexes (2 levels) and each latitudinal group (3 levels) using two factor ANOVA. One-way ANOVAs were performed on Female and Male birds separately where the sex x latitude group interaction was significant (P <0.05) to determine if mean values varied significantly between latitude groups. Mean values for male and female birds are given separately. Values in brackets are standard errors.

Replacement ANOVA p values Av. Morphometrics Values§ Variable 1st Factor 2nd Factor Mean N SEX † Lat Group¶ Interaction Male Female

Weight (g) - 57 0.0001 0.0005 0.7537 2651.1(91.5) 3663.2(113.3) Wing cord (mm) 588.1 90 0.0001 0.0001 0.6643 577.8(4.3) 610.2(4.2) top of the wrist to the tip of the longest primary Head length (mm) 111.9 87 0.0001 0.0178 0.3709 110.1(0.7) 115.8(39.0) back of the head to the tip of the bill Tail length (mm) 307.3 89 0.044 0.0001 0.3273 311.1(4.7) 323.9(5.2) Culmen tip/cere (mm) 42.6 86 0.0001 0.3511 0.5711 40.2(0.5) 44.9(0.7) tip of the bill to the front edge of the cere Culmen tip/base (mm) 51.9 87 0.0001 0.7882 0.6595 50.5(0.4) 55.1(0.7) tip of the bill to the plumage edge at the front of the head Culmen width (mm) - 20 0.1912 0.852 0.7790 20.5(0.7) 22.46(0.8) width at the cere Cont’d overleaf

108

Cont’d from previous page Variable 1st Factor 2nd Factor Mean N SEX † Lat Group¶ Interaction Male Female Culmen depth (mm) - 20 0.1005 0.0812 0.7701 26.6(0.9) 28.8(0.9) depth at the cere Tarsal joint (mm) 23.3 85 0.0001 0.0038 0.7332 22.1(0.2) 24.4(0.3) Tarsal length (mm) 94.9 90 0.0034 0.0367 0.4421 95.1(1.0) 100.2(1.4) Tarsal breadth (mm) 16.8 89 0.0002 0.0029 0.2504 15.7(0.3) 17.5(0.3) taken at the midpoint Tarsal width (mm) 16.2 89 0.0401 0.3425 0.3380 14.4(0.2) 20.1(3.0) taken at the midpoint Halux claw length (HxL, mm) 38.2 89 0.0001 0.4107 0.0121 35.8(0.3) 40.5(0.3) HxL - Female 41 0.0107 HxL - Male 48 0.3658 Halux claw breadth (mm) 9.5 89 0.0001 0.1855 0.1520 9.1(0.1) 10.1(0.2) depth of the claw at the intersection with the toe Halux claw width (mm) 7.2 88 0.0005 0.5810 0.2976 6.8(0.1) 7.3(0.1) width of the claw at the intersection with the toe † Male, Female ¶ North, Middle, South § “-“ indicates that this variable was excluded from the principal component analysis

109 Chapter 5 Morphometric Variation

Second, individuals were assigned to regions previously used to quantify genetic structure in this species (see Chapter 2) (Fig. 5.1). Principal component analysis

(SAS, 1996) based on a correlation matrix (Quinn and Keough, 2002) was used to determine if the morphometric data grouped naturally to these regions. Male and female data were analysed separately. Prior to multivariate analysis, missing data were replaced with an average given in Table 5.1 (calculated from data in Chapter 4).

Each morphometric variable was checked for deviation from a normal distribution using normal probability plots and measures of skewness and kurtosis (Tabachnick and Fidell, 2001). Width at cere and depth at cere were dropped from the analysis due to insufficient data. As in the sex discrimination analysis, weight was not included in the analysis as it was considered unreliable due to the inclusion of wild and captive individuals, and may reflect condition or seasonal variation (Smith and Wiemeyer,

1992). Therefore, the principal component analysis included 12 of the original morphometric characters (Table 5.1). Amongst these 12 characters, the level of replacement data ranged from 2.1% - 7.4%. 19% of individuals had missing data.

Four individuals were removed from this analysis due to an unacceptably high number of missing values, leaving 42 females and 49 males.

The relationship between scores from the first principal component (Prin 1) and latitude and longitude was tested using least squares linear regression (SAS, 1996).

To determine if there were regionalised patterns of morphometric variation, the distribution of mean scores from Prin 1 were compared among regions (Fig. 5.1) using single factor ANOVA, for males and females separately. All pairwise

110 Chapter 5 Morphometric Variation

## #

###7

# ## # # # # # # # #

## # # # # 4 # # # #

# ## # 5 3 ## # #

6 # # # 2 # #

# ## # # #

# # # # ### # ## ## ## # ### 1 ### ######

Figure 5.1 Map showing the source of samples relative to regions used to quantify genetic structure (Chapter 2) in H. leucogaster. These are the same regions used to calculate morphometric and genetic distance. Regions are: 1) Tasmania, 2) Southeast Australia, 3) Central and Southeast Queensland, 4) Northern Australia, 5) Western Australia, 6) South Australia, 7) Papua New Guinea.

differences among mean values were tested with LSD comparisons using the “GLM” and “lsmeans” procedures in SAS (1996).

111 Chapter 5 Morphometric Variation

5.2.2 Test of genetic and morphometric relationship

A Mantel test (1000 randomisations; Belbin, 1994) using morphometric distance and genetic distance (Slatkin’s linearised ΦST) between regions (Fig. 5.1) was used to investigate whether pairs of regions that were more divergent morphologically were also more different genetically. An average was calculated for each morphometric character within each region for male (N=49) and female (N=42) data. Morphometric distance was then calculated based on Euclidean distance in PATN (Belbin, 1994).

Region 6 (South Australia) could not be included in the male test due to insufficient data. A pairwise matrix was calculated for each sex (Table 5.2). Genetic distance was calculated in ARLEQUIN (Schneider et al., 2000) using mtDNA control region sequence data from Chapter 2. The same genetic matrix was used for each Mantel test, excluding Region 6 from the male test (Table 5.2). It was not possible to use identical data to calculate the morphometric and genetic distance matrices as morphometric data was not available for all genetic samples. 55 out of 91 individuals

(60.4%) were common to both data sets (Table 5.3). The genetic distance matrix data was based on data from 125 individuals (Chapter 2).

112

Female Tasmania Southeast Aust Central & SE Qld Northern Aust Western Aust South Aust PNG

Tasmania - 4.21 3.80 8.01 6.29 3.17 5.25 Southeast Australia 0.00310 - 1.27 4.62 7.86 6.40 1.30 Central & Southeast Qld 0.00479 0.06010 - 5.09 7.43 5.59 2.27 Northern Australia 0.00000 0.04458 0.02302 - 8.54 10.34 4.32 Western Australia 0.03081 0.08645 0.00000 0.01226 - 7.48 8.92 South Australia 0.01365 0.14355 0.00000 0.06990 0.06258 - 7.46 Papua New Guinea 0.05089 0.85227 0.00000 0.14150 0.00000 0.05971 -

Male Tasmania - 3.64 3.22 8.90 4.33 10.38 Southeast Australia 0.00310 - 0.93 6.12 2.03 8.42 Central & Southeast Qld 0.00479 0.06010 - 6.16 1.75 8.24 Northern Australia 0.00000 0.04458 0.02302 - 4.63 3.25 Western Australia 0.03081 0.08645 0.00000 0.01226 - 6.60 Papua New Guinea 0.05089 0.85227 0.00000 0.14150 0.00000 -

113 Chapter 5 Morphometric Variation

Region Male Female Shared

Tasmania 19 19 24 Southeast Aust 7 3 6 Central & Southeast Queensland 7 7 8 Northern Australia 8 8 9 Western Australia 3 1 3 South Australia 0 1 1 Papua New Guinea 5 3 4

Total 49 42 55

A second series of Mantel tests were run using the same male and female morphometric matrices to determine if morphometric similarity between regions was correlated with geographic proximity. It is not known whether long distance dispersal in H. leucogaster is restricted to coastal or riparian movement in favour of direct overland routes, so both male and female morphometric matrices were compared against coastline and straightline distance. Distance was calculated between regional centroids using a GIS (ESRI, 1996).

114 Chapter 5 Morphometric Variation

5.3 Results

Eight out of 15 morphometric variables showed differences in mean size between

Latitude Groups. In all cases where significant differences were found (Table 5.1), the

Southern group differed from at least one of the other groups (Fig. 5.2). Mean wing chord and tail length differed significantly between all Latitude Groups. Variation in halux claw length between the sexes was not independent of the effect of latitude

(Table 5.1), and was only significantly different between female groups, with largest claw lengths in the South (Fig 5.2).

There is some suggestion in the data of Marchant and Higgins (1993) that sub-adults show reduced dimorphism between sexes. The effect of this in the current data would have been to reduce mean variation between the sexes. There was no effect of latitude on morphological variation between adult and sub-adult birds. Both male (two way

ANOVA; F=20.69, d.f.=1, P=0.001, N=47) and female (two way ANOVA; F=4.67, d.f.=1, P=0.038, N=41) sub-adults showed significantly longer tails than their adult counterparts. Other than tail length, age-related size differences were indiscernible, so geographical size differences in sub-adults are considered similar to those in adults.

In Chapter 4 I found that the inclusion of museum skins potentially biased some measurements. In all cases these measurements described tarsal bone characteristics, with mean measurements smaller among museum skins than other sample types, probably due to the shrinkage of tissue around the bone. Although 64% of female and

68% of male measurements were derived from museum skins, samples were

115 Chapter 5 Morphometric Variation

4500 640 B C 4000 Female A A 620 B 3500 A Male ) 600 3000 m m ) ( g d ( 2500 580 r t o h

ig 2000 560 e ng C W 1500 i

W 540 1000 500 520 0 500 North Mid South North Mid South

400 120 B C 30 B B 118 350 A A A A AB 25 116 ) ) ) ) ) 300 ) m m m m m 114 m m m m m m 20 (m ( ( ( 250 ( ( t t h h 112 h th t n n i i g n ng 200 ngt 110 ngt 15 e Jo Jo Le L 108 150 sal sal d il Le il Le r r ad a a a 10 a a e e 106 T

100 T H 104 5 102 50 100 0 0 North Mid South North Mid South North Mid South

110 20 B 45 aa b B B 18 A 40 105 ) m ) ) ) ) 16

m 35 m A m m m A ( m m m m 14 h 100 t 30 h ( h ( g t t

12 n d d

e 25 a a ngth ( ngth ( e e 95 10 L r r

w 20 Le Le a B B l l 8 a a s s 90 Cl 15 sal sal r r x r r 6 a a a a u l T T 4 10

85 Ha 2 5 80 0 0 North Mid South North Mid South North Mid South

Figure 5.2 Male and female measurements (mean ± s.e.) for each morphometric variable in which there was significant variation between latitude groups. Regions with the same letter are not significantly different (P >0.05). Lower case letters show results of single factor ANOVA for the female group only, where the interaction term (Sex x Latitude Group) was significant.

116 Chapter 5 Morphometric Variation

distributed evenly among Regions and Latitude Groups in both sexes. Therefore, any bias would have been equally applied, and was unlikely to have affected either the

ANOVA or principal component results.

In both sexes, the first principal component scores from the multivariate morphometric analysis were correlated with latitude (female: r2=0.26, P=0.0006; male: r2=0.23, P=0.0006). Some effect of longitude was also found on the female sample (female: r2 =0.12, P=0.03; male: r2 = 0.009, P=0.51). Individuals clustered loosely within regions, with regions from similar latitudes grouping together (Fig. 5.3

& 5.4). Clearest separation was seen among regions from the North Latitude Group in relation to the South Latitude Group (Fig. 5.3 & 5.4). Female birds from Tasmania differed significantly from those from Northern Australia (P=0.01) and Papua New

Guinea (P= 0.03) (Fig. 5.5a), although mean differences in first principal component scores were not significantly different when considering all regional comparisons

(one-way ANOVA; F=2.20, d.f. = 6,41 P = 0.07). Similar latitudinal differences occurred for males, and regions showed significant morphological differences (one- way ANOVA; F=3.85, d.f. = 5,48 P = 0.006) (Fig. 5.5b), although the strongest differences were between the north and all other regions. Some evidence for an effect of longitude in the female sample may come from the very different first principal component scores between Tasmania and Western Australia (Fig. 5.5a); though these regions were not statistically different in the ANOVA comparison.

Among females, morphometric similarity was significantly correlated with coastline distance suggesting that birds geographically close are morphometrically more

117 Chapter 5 Morphometric Variation

a) PRIN 2 North Mid South

PRIN 1

b) PRIN 2 Tasmania

Southeast Aust

Central & SE Queensland Northern Aust

Western Aust PRIN 1 South Aust

Papua New Guinea

Figure 5.3 Distribution of Principal Component scores for female H. leucogaster relative to a) latitude groups, and b) regions.

118 Chapter 5 Morphometric Variation

a) PRIN 2 North Mid South

PRIN 1

PRIN 2 Tasmania

Southeast Aust

Central & SE Queensland Northern Aust

PRIN 1 Western Aust

Papua New Guinea

Figure 5.4 Distribution of Principal Component scores for male H. leucogaster relative to a) latitude groups, and b) regions.

119 Chapter 5 Morphometric Variation

a) b) 3 3 A BC 2 2 A A AB 1 1 C 0 0 -1 -1 -2 -2 -3 -3

-4 t d a a s st l i t t t l d a a a u u l s s s i i nea a l Q i e r u u ani n t A a A n u Q r a s i t st Au A rn a t s u st sm m h A G ast e s rn u E a s ut w a

e Ea T o e he n A w Gu A e rth Ta r n h o rth S t r e S. o u out N e l & S. Ne N & S a l a ua N r So u a t p r p West n t a West n a P P Ce Ce

Figure 5.5 Principal component scores (mean ± s.e.) for a) female and b) male birds for each region. Regions with the same letter are not significantly different (P >0.05).

similar (Table 5.4). None of the other tests showed any correlation between

morphometric distance and geographic distance (Table 5.4). Similarly, genetic

differentiation and morphometric variation between regions was not significantly

related in either males or females (Table 5.4).

Within sexes, the first five Principal Components explained 66% and 67% of the

variation among female and male birds respectively. Base on the first principal

component, variation in claw structure and wing chord most clearly separated

120 Chapter 5 Morphometric Variation

females, whereas a combination of wing chord, tail and culmen length and lower leg structure appeared to separate male birds (Table 5.5).

Table 5.4 Relationship between morphometric distance and either geographic or genetic distance using Mantel’s Test. Bold values are significant (P <0.01).

Males Females Comparison r P r P Morph distance vs coastal distance 0.207 0.236 0.575 0.0004 Morph distance vs straightline distance 0.353 0.086 0.271 0.115 Morph distance vs genetic distance 0.25 0.226 -0.376 0.933

Table 5.5 Loadings for the first and second Principal Components for each sex.

Female Male

Variable Prin 1 Prin 2 Prin 1 Prin 2 Wing cord 0.58 0.01 0.67 -0.30 Head length 0.47 -0.11 0.49 -0.11 Tail length 0.41 -0.51 0.53 -0.51 Culmen tip to cere 0.31 0.62 0.54 0.28 Culmen tip to base 0.43 0.54 0.37 0.43 Tarsal joint 0.32 -0.32 0.33 -0.68 Tarsal length 0.02 0.20 0.21 0.55 Tarsal width 0.36 0.42 0.50 -0.23 Tarsal breadth 0.39 -0.44 0.51 0.17 Halux claw length 0.76 -0.10 0.47 0.32 Halux claw width 0.49 0.22 0.43 0.47 Halux breadth 0.79 -0.11 0.26 0.03 Eigenvalue 2.84 1.54 2.54 1.80 % variation explained 23.6 12.8 21.2 15.0

121 Chapter 5 Morphometric Variation

5.4 Discussion

Given the lack of genetic structure in this species (Chapter 2), a similar lack of morphometric structure may have been expected. However, there appears to be clear separation in the morphology of birds at least between the southern latitudes and all other regions. There is also some evidence of an isolation by distance effect among females that was not found among males, where birds that were geographically closer, were also morphologically more similar. This is interesting as there is no isolation by distance effect based on genetic similarlity (Chapter 2), nor is there a relationship between morphological similarity and genetic similarity.

There are three possible explanations for the maintenance of morphometric variation without evidence for genetic structure in this species. First, morphometric variation may be a result of different conditions within local environments and may reflect some form of phenotypically plastic response among individuals from different regions. Second, current selection pressure may be maintaining predictable morphological variation in the face of gene flow as measured by the neutral variation in mtDNA. Third, historically isolated populations established through long-range dispersal following the original colonisation may have merged, through secondary contact, to produce the current distribution. In the case of phenotypic plasticity or current selection, you would not expect correspondence between patterns of morphological and genetic variation. However, if colonisation history has affected the current level of genetic variation between regions, or if some areas are genetically distinct, this may be reflected in the pattern of morphometric similarity between regions.

122 Chapter 5 Morphometric Variation

There is little support for secondary contact (Endler, 1977) producing latitudinal division in size variation in this species as it would have been reflected in the existing genetic data (Chapter 2) which shows no significant genetic structuring within the

Australo-Papuan region. However, it is possible that there may be less gene flow in this species than originally suggested. For example, in Chapter 2 it was estimated that there has only been approximately 32 000 generations since colonisation, with the current distribution having grown from a relatively small initial population. It is possible that there has been insufficient time to reach an equilibrium between gene flow and mutation (Wright, 1951). In this instance, at least some of the apparent movement of individuals between regions may be a result of population expansion rather than current gene flow (Greenberg et al., 1998).

Studies have also shown that populations which have expanded from a bottleneck typically show low haplotype and nucleotide diversity (e.g.Glenn et al., 1999; Mila et al., 2000). This may be interpreted as a lack of genetic structure due to ongoing gene flow, but is equally indicative of historical processes in the absence of current movement (Zink, 1997).

Additionally, Greenberg et al. (1998) argued that polygenic traits responsible for maintaining morphometric variation may have very high mutation rates. Following a bottleneck, variation may recover more quickly among these traits than for mtDNA which is effectively only a single locus. Similar arguments have been put forward to suggest that morphometric variation may have evolved over a time scale that is too short to be detectable by mtDNA assays (Ball et al., 1988; Seutin et al., 1995; Storfer,

123 Chapter 5 Morphometric Variation

1996; Grapputo et al., 1998). Therefore, it is possible that the colonisation history of this species, or the inability of the genetic marker to reveal recent evolutionary subdivision, may be masking a higher level of population structure.

Even given the potential for some restriction to gene flow, there is no doubt that H. leucogaster is capable of long distance dispersal. Mark-recapture records in this species are limited, but a juvenile is known to have moved 3000 km from its natal territory in South Australia to Fraser Island in Queensland (Marchant and Higgins,

1993). Similarly, analysis in Chapter 3 on the distribution and occupancy of H. leucogaster within Australia from 1901 to 2001, showed that while the spatial extent of occupancy has remained relatively stable, there has been considerable change in the degree of occupancy and range extent at the regional level. The over-riding factor contributing to these changes was medium-term climate fluctuation (Chapter 3).

Although it is known that offspring are driven from parental territories (Marchant and

Higgins, 1993), there has been no direct test of actual dispersal distance in this species. So it is unclear how far juvenile birds may move before establishing a territory of their own. Distance travelled prior to recruitment may be smaller than anticipated, remembering that under theoretical expectations only a single individual is required to move between regions per generation to prevent divergence to genetic fixation (Avise, 1994; Zink et al., 2000). If recruitment distances were shorter than expected, this could lead to greater structure. The only evidence for this in the current data comes from the signature of isolation by distance among female birds based on morphometric variation between regions. This suggests that female birds may not

124 Chapter 5 Morphometric Variation

move as far from natal territories as males. It is not possible to examine this hypothesis using mtDNA data, but the potential for sex-biased dispersal could be addressed with a nuclear marker. An alternate explanation is that there are differential selection pressures acting on male and female birds.

Latitudinal variation may also be due to phenotypic plasticity or selection. In the case of phenotypic plasticity, variation is due to the impact of environmental factors, rather than an evolutionary response to some selective force, and genetic differences between generations will remain constant given an unchanged environment (Smith-

Gill, 1983; Futuyma, 1986), producing continuous phenotypic variation in response to environmental factors (Swain and Foote, 1999). For example, populations of the Red- winged Blackbird Agelaius phoeniceus show a similar pattern of genetic homogeneity to H. leucogaster, together with stable clines of morphological character variation across their range (James, 1983). In carefully manipulated translocation experiments, young reared away from their natal origin displayed phenotypic characteristics of the new site. In a similar experiment, manipulation of juvenile brook charr Salvelinus frontalis showed differential caudal fin growth due to changed water velocity, with individuals from low-velocity sites developing larger caudal fins when reared in a high-velocity treatment (Imre et al., 2002). While there may be some effect of phenotypic modulation in H. leucogaster, without similar controlled experiments among populations it is not possible to confirm any role of phenotypic plasticity on the maintenance of size variation.

125 Chapter 5 Morphometric Variation

The most commonly cited explanation for size variation along a latitudinal cline is

Bergmann’s Rule which states that continuously distributed species tend to be larger in regions further from the equator (Zink and Remsen, 1986). This pattern is attributed to thermoregulatory needs in colder climates where a low surface area to volume ratio may be advantageous (Fitzpatrick and Dunk, 1999). Differences in breeding phenology between northern and southern populations of H. leucogaster

(Marchant and Higgins, 1993) suggest temperature may play a significant role in some aspects of this species’ biology. However, reviews of morphometric variation among bird species have shown compliance with the body-size hypothesis in as many cases as there has been disagreement (Zink and Remsen, 1986; Fitzpatrick and Dunk,

1999), and it has been suggested that accepting temperature variation as a singular selective pressure may be misplaced (Begon et al., 1990). An alternate explanation is that the link between morphology and latitude may be an indirect consequence of adaptation to prey type and availability rather than climate per se (Fitzpatrick and

Dunk, 1999). Similarly, Zink and Remsen (1986) hypothesised that colder climates may dictate larger territories requiring more flights in a day to provide sufficient resources, therefore favouring larger individuals with longer wings. Wing length, halux claw size and tarsal characteristics were defined in the principal component analysis (Table 5.5) as determining size variation in female and male birds in this study, and may be related to feeding ecology within different latitudinal zones.

According to Endler (1986), the most likely explanation for a stable cline of character variation is selection. That is, without selection this geographic pattern is less likely than the stability of character traits in only one or two locations. Therefore, it seems reasonable to suggest that some form of selection may be acting on these characters.

126 Chapter 5 Morphometric Variation

Accordingly, morphometric variation in populations of H. leucogaster could be explained by thermoregulatory requirements, adaptation to prey type or environmentally induced plasticity. While additional experimental research would be required to confirm that morphometric variation was a result of environmental induction, the more likely explanation is that while moderate levels of dispersal among regions are maintaining a relatively homogenous genetic population structure, it is not sufficient to swamp the effects of selection.

127 Chapter 6 General Discussion

Chapter 6 – General Discussion

A species distribution and consequent level of population structure will be determined by a range of factors including, but not limited to: colonisation history, dispersal, behaviour, environmental stochasticity and habitat heterogeneity (Zink, 1997;

Gandon and Michalakos, 2001). The ability of direct methods of ecological survey to explore these mechanisms has been shown to be limited as surveys are usually temporally and spatially restricted, often to scales smaller than a species’ known geographic range (Goodwin and Fahrig, 1998). For example, in a review by Marzluff and Sallabanks (1998b), of 159 studies conducted between 1990 and 1996, 73% lasted less than three years and only 5% lasted ten or more years. Similarly, direct measures of dispersal often do not agree with those inferred from gene flow studies and may be prohibitive in terms of the time it takes to collect data or feasibly track species long distance movements (Baker et al., 2001; Peacock and Ray, 2001).

The work in this thesis was motivated by a lack of knowledge regarding the current population structure and status of H. leucogaster in Australia, as well as a need to identify evidence of population decline or shifts in range extent. Prior assessments of conservation status in this species (see Chapter 1) have been based on regional or local level observations, and on the whole in the absence of ecological survey or testing.

To remedy this, the combined indirect methodologies applied in this study circumvent the restrictions imposed by direct ecological sampling described above, by allowing

128 Chapter 6 General Discussion

survey across large geographic and temporal scales, which effectively covered the entire Australian range of H. leucogaster. These methods also allowed a consideration of evolutionary factors likely to underpin current patterns of distribution.

The focus of this chapter is to provide a synthesis of information from the preceding chapters to recommend an appropriate geographic scale at which to monitor population change, and having adopted that scale, to determine if, based on the current data, there is statistical evidence of decline in H. leucogaster. Finally, some outstanding issues and future research directions are outlined.

6.1 Monitoring population fluctuation: the issue of scale

In a realistic evaluation of the likelihood of species population persistence, the choice of scale is paramount. Ecological systems are expected to show variability across a range of temporal and spatial scales (Levin, 1992), with observed patterns dependent on the scale of analysis. Therefore it is possible to misrepresent patterns, or change to patterns of distribution, if the scale of data collection and analysis is not reconciled with the spatial scale of the study question.

Accordingly, two over-arching themes have run through the preceding chapters. First, is there significant variation within the parameters measured or characterised at different spatial or temporal scales in this study? Second, given this information, at what physical scale is it appropriate to monitor change in population dynamics in the

White-bellied Sea-Eagle?

129 Chapter 6 General Discussion

In Chapter 1 I suggested that the scale at which demographic trends should be analysed may be difficult to determine if population boundaries are unclear.

Specifically among avian species, the definition of population boundaries can be complicated by large-scale distributions, or peculiarities of social or kin relations.

Hence, an indirect technique such as DNA analysis may be useful in resolving such issues.

As might be expected given the physical scale over which this study was conducted, and given the theoretical considerations of Levin (1992), the pattern of genetic, morphometric and physical distribution in this study varied dependent on the scale of analysis. Regional patterns of genetic variation (Chapter 2), trends in occupancy and density (Chapter 3), and regional morphometric distributions (Chapter 5) were not representative of continental patterns, reinforcing the contention that the extrapolation of data from local or regional levels is often inappropriate (Flather and Sauer, 1996;

Maurer and Villard, 1996; Turner et al., 2001). For example, Western Australia and

South Australia each contained four mitochondrial DNA haplotypes, but shared only one (Table 2.4). Overall, other regional differences in haplotype distribution existed but were not statistically different at any geographic scale of comparison. Similarly, an index of the frequency of individuals within regional one-degree grid cells was several orders of magnitude larger in north Queensland than in South Australia, but showed a trend of decline in north Queensland, while occupancy in South Australia remained relatively stable across the same period (Fig 3.5). Again, although there were marked changes in frequency and range extent between defined regions, this had

130 Chapter 6 General Discussion

no bearing on comparisons at the continental scale between the periods 1901- 1976,

1977-1981 and 1998-2001, which were not significantly different.

The lack of genetic population structure within Australia (Chapter 2) implies that the ecological survey of population fluctuation or change is appropriate at the continental scale. When viewing results of the spatial analyses of historical changes in distribution (Chapter 3), this conclusion is supported since at a continent-wide scale the average frequency of the species within one-degree grid cells did not vary significantly between 1901 and 2001.

Therefore, based on the data from Chapter 2 and 3, it is a recommendation of this study that future population monitoring is appropriate at the continental scale.

6.2 Evidence of population decline from the data?

At the level of the individual, the first sign of long-term directional change in abundance may be density declines within occupied blocks at large ecological scales

(Gaston et al., 2000). At the genetic level, reduced genetic variability has been linked to inbreeding and associated fitness problems, and may reduce a species ability to face future environmental challenges (O'Brien et al., 1996).

Based on analysis in Chapter 3, at the continental scale and across the complete timeframe (1901-2001), there is little evidence to support claims of persistent decline in the abundance of H. leucogaster. However, had the analysis been restricted to a comparison of frequency between only the Atlas 1 (1977-1981) and Atlas 2 (1998-

131 Chapter 6 General Discussion

2001) periods, it is possible to see how a public perception of population decline may have originated. For example, among high frequency coastal blocks there were non- significant reductions in frequency in conjunction with obvious coastal land-use change, which may have reinforced a perception of overall population decline.

From a genetic perspective, both mtDNA nucleotide and haplotype diversities were low in this study (Table 2.3). Haplotype diversity ranged from zero in Southeast

Australia to 0.574 ± 0.219 in Central and Southeast Queensland, with an average diversity within Australia of 0.314 ± 0.055. These values are considered low in comparison with other bird species (Vallianatos et al., 2002). Importantly, low genetic variabilities are not necessarily indicative of reduced viability as they may represent the outcome, rather than the cause, of population size reduction (Haig and

Avise, 1996), as may occur when only a small number of individuals colonise an area.

Indeed Chapter 2 concluded that H. leucogaster suffered a bottleneck at the time that the species colonised Australia. Moreover, new haplotypes appear to have arisen since colonisation (Chapter 2). Since there is no evidence for actual physical decline in occupancy over ecological time scales (Chapter 3), low haplotype diversity in this species is more likely a reflection of colonisation history rather than reduced fitness.

It is unclear what factors may be driving the genetic homogeneity within south eastern Australia. This result may imply some form of isolation or regionalised contemporary bottleneck. Inland blocks in particular showed declines in the frequency of H. leucogaster sightings, although these were not statistically significant

(Fig 3.5). There has been a reduction in forested land in south eastern Australia

132 Chapter 6 General Discussion

(Barson et al., 2000), and clearing of coastal forest due to agricultural and urban expansion has been linked to the direct loss of H. leucogaster nests and reduced breeding success locally within Victoria (Clunie, 1994). It seems possible that nest site limitation in this region could prevent recruitment from outside the immediate gene pool, although this would require direct testing to disprove.

Finally, a number of authors have suggested that demographic response lags may mask persistent levels of decline. Significant response lags in declining woodland bird species have been suggested for the New South Wales sheep-wheat belt following vegetation clearing, with declines continuing long after clearing had ceased

(Reid, 2000). Chamberlain et al. (2000) suggested that conservation recommendations made only a few years after land use change are most likely to be invalid due to population response lags. Given H. leucogaster’s long generation time

(approximately five years), it is possible that the demographic impact of habitat change at the landscape scale within the south eastern Australia has not yet been fully realised. In this instance, even longer-term data sets may be required to reveal significant levels of potential future decline. Alternatively, the current level of genetic homogeneity may be an interim response to recent habitat change prior to a return to previous levels of genetic diversity.

While additional analysis may be warranted in south east Australia to clarify the above, across most of the continent and based on the current genetic and distributional data, there is limited evidence to suggest that H. leucogaster is declining. However, given the lack of diversity in south eastern Australia, the

133 Chapter 6 General Discussion

indication of recent population expansion and the clear maintenance of morphometric differentiation among latitude groups (Chapter 5), some caution should be exercised in accepting that population monitoring is singularly appropriate at the continental scale. That is, if an alternate pattern of population structure exists that has not been discernible with the existing analysis, the recommended scale of population monitoring may require review.

6.3 Evidence for an alternate pattern of population structure

Evidence based on stable clines of morphometric variation (Chapter 5) and incomplete sharing of haplotypes between the west and east of the continent (Chapter

2) suggests there may be less contemporary interbreeding of birds across the continent than was suggested by the mtDNA analysis. From a management perspective the consequence of this may be that population monitoring at the continental scale is too large to effectively detect important regional level changes.

In the context of this study, dispersal and therefore the potential for genetic exchange or change in distributional patterns, is likely to occur under two conditions. First, due to natal dispersal, defined here as the distance travelled between birth and establishing a permanent breeding territory. Second, in response to some form of disturbance.

Based on the distributional data (Chapter 3) it is not possible to say whether bird movement in response to drought conditions was between regions or restricted within regions. Considering the potential costs of dispersal (e.g. mortality; Gandon and

Michalakis, 2001), it is likely that birds would move the minimum distance necessary

134 Chapter 6 General Discussion

to escape unsuitable environmental conditions or to establish a breeding territory.

Therefore inter-regional dispersal may not be as widespread as predicted.

Additionally, for selection to maintain the stable cline of morphometric variation suggested in Chapter 5 there must be some restriction to gene flow (Barton, 2001).

Evidence of shorter recruitment distances is also suggested by the signature of isolation by distance among female birds based on morphometric variation between regions (Chapter 5). This suggested that female birds may not move as far from natal territories as male birds. One possible explanation for this is that genes effecting morphometric traits driving this pattern are located on the ‘W’ chromosome. As females are the heterogametic sex in birds this would affect females only.

Most avian species display female sex-biased dispersal, although there are exceptions such as the Lesser Snow Goose (Chen caerulescens) and the Australian Magpie

(Gymnorhina tibicen) in which inter-colony gene flow is mediated by males (Avise,

1994; Veltman and Carrick, 1990). To test for male-biased dispersal, complementary analysis with nuclear markers is required. For male-biased dispersal to be confirmed, there would need to be a reduced signature of maternal gene flow relative to nuclear gene flow. In this instance, population monitoring would still be appropriate at the continental scale.

Another important reason to employ complementary analysis using a nuclear marker is to confirm that the Australian population was subject to a bottleneck at the time of colonisation. To confirm this, it is necessary to rule out the alternate hypothesis that

135 Chapter 6 General Discussion

the mitochondrial genome has been subjected to a ‘selective sweep’ (Rand, 1996). In this scenario all descendant alleles pass through the ancestry of a favoured allele

(Rand, 1996; Avise, 2000), producing a reduction in mitochondrial diversity that is not replicated among independent nuclear loci. A true demographic bottleneck will affect all loci equally (Rand, 1996), and complementary nuclear loci should show the same relative levels of reduced diversity as the mtDNA analysis.

Interestingly, the Carpentaria break and Nullarbor Plain have been identified as barriers to dispersal among some species of Australian raptor (Olsen, 1995; Fig.

2.2b). Given the inherent dispersal ability of H. leucogaster, it seems unlikely that they would function as effective barriers to dispersal. However, taken the absence of

Haplotypes 7, 8 and 12 in the east (Chapter 2; Fig. 2.4), additional samples from the north and west of the continent are required to rule out some restriction of maternal gene flow between the east and west of the continent.

Mitochondrial DNA has been the marker of choice in conservation genetic studies

(Avise, 1995). However, conservation decisions made on the basis of mtDNA analysis alone have been criticised, as population structuring is not always replicated with nuclear markers (Moritz, 1994a; Avise, 1995; Avise, 2000). While alternative patterns of population structure may be revealed using unlinked nuclear markers

(given that a bottleneck is confirmed), the larger effective population size of nuclear

DNA and slower time to reach equilibrium (Crochet, 2000), should produce even lower levels of genetic homogeneity among H. leucogaster populations than suggested in the mtDNA analysis. In this case, the finding of no significant level of

136 Chapter 6 General Discussion

population structure would be verified, and again population monitoring would be appropriate at the continental scale.

In the event that there was significant evidence for natal philopatry among female birds, contemporary levels of mixing and recolonisation potential may have been overestimated. While the current level of genetic exchange between regions is clouded by the finding of a recent colonisation event, even occasional long-range movement between regions has the potential to produce genetic homogeneity as well as recolonise habitat patches that have been subject to stochastic loss (Nichols and

Hewitt, 1994) suggesting that continent scale population monitoring is still appropriate.

To clarify this, priority in future research should be given to ruling out the possibility of a selective sweep through analyses of nuclear DNA. Furthermore, additional geographic sampling should be done to ensure that there is not a greater level of mtDNA population structure than suggested by the current analysis. That is, small sample size or inadequate sample distribution should be ruled out as being responsible for the current pattern of genetic structure.

6.4 Expansion of genetic analysis outside the Australian range of H.

leucogaster

Expansion of the genetic analysis to encompass the remainder of the range of H. leucogaster would provide opportunities to address a number of questions. First, there is no information about the evolutionary relationship between the Indian and

137 Chapter 6 General Discussion

Southeast Asian populations. Inclusion of samples from these regions within the existing analysis would allow the identification of the centre of the species distribution as well as characterisation of the complete phylogeographic tree

(Crandall and Templeton, 1993). This would yield interesting insights into the source region of the species radiation and possibly the timing of its expansion. Additionally,

Singapore haplotypes differed from the Australian haplotypes by only three base pairs. This suggests not only that clade 1-1 haplotypes may be part of the Southeast

Asian distribution, but that the overall evolutionary tree may be quite shallow.

Surveying additional haplotypes from Southeast Asia may also confirm whether colonisation of Australia was through Timor as well as Papua New Guinea. This would also allow estimates of the relative isolation of the current Australian distribution of H. leucogaster from the remainder of its range, which may have significant future conservation implications.

138 References

References

AUSLIG (2001) Global Map Data Australia 1M 2001 Australian Surveying and Land

Information Group (AUSLIG).

Allen GR (1989) Freshwater fishes of Australia. TFH Publications, Brookvale, NSW.

Aspinall R (1992) Spatial Analysis of Wildlife Distribution and Habitat In a GIS. The

Proceedings of the 5th International Symposium on Spatial Data Handling.

IGU Commision on GIS, Aug 3-7 1992., 2, 444-453.

Avise JC, Arnold J, Ball RM, Bermingham E, Lamb T, Neigel JE, Reeb J, Saunders

NC (1987) Intraspecific phylogeography: the mitochondrial DNA bridge

between population genetics and systematics. Annual Review in Ecology and

Systematics, 18, 498-522.

Avise JC (1994) Molecular Markers, Natural History and Evolution. Chapman &

Hall, N.Y.

Avise JC (1995) Mitochondrial DNA polymorphism and a connection between

genetics and demography of relevance to conservation. Conservation Biology,

9(3), 686-690.

Avise JC (1996) Three fundamental contributions of molecular genetics to avian

ecology and evolution. Ibis, 138, 16-25.

Avise JC, Walker D (1998) Pleistocene phylogeographic effects on avian populations

and the speciation process. Proceedings of the Royal Society of London. Series

B, 265, 457-463.

Avise JC (2000) Phylogeography: The History and Formation of Species. Harvard

University Press, Cambridge, Massachusetts.

139 References

Baker AJ, Marshall HD (1997) Mitochondrial control region sequences as tools for

understanding evolution. In: Avian Molecular Evolution and Systematics. (eds.

Mindell DP), pp. 51-82 Academic Press, USA.

Baker AM, Mather PB, Hughes JM (2001) Evidence for long-distance dispersal in a

sedentary passerine, Gymnorhina tibicen (Artamidae). Biological Journal of

the Linnean Society, 72, 333-343.

Baker-Gabb DJ (1984) Morphometric data and dimorphism indices of some

Australian raptors. Corella, 8 (3), 61-63.

Balbontin J, Ferrer M, Casado E (2001) Sex determination in Booted Eagles

(Hieraaetus pennatus) using molecular procedures and discriminant function

analysis. Journal of Raptor Research, 35 (1), 20-23.

Ball RM, Freeman S, James FC, Bermingham E (1988) Phylogeographic population

structure of Red-Winged Blackbirds assessed by mitochondrial DNA.

Proceedings of the National Academy of Sciences USA, 85, 1558-1562.

Barker F, Barrowclough GF, Groth JG (2002) A phylogenetic hypothesis for

passerine birds: taxonomic and biogeographic implications of an analysis of

nuclear DNA sequence data. Proceedings of the Royal Society of London.

Series B, 269, 295-308.

Barker RJ, Sauer JR (1992) Modelling population change from time series data. In:

Wildlife 2001: populations. (eds. McCullough DR, and Barrett RH), pp. 182-

194 Elsevier Applied Science, New York.

Barrett G, Silcocks A, Barry S, Poulter R, Cunningham R (2003) The New Atlas Of

Australian Birds. Royal Australasian Ornithologists Union, Melbourne.

140 References

Barson M, Randall L, Bordas V (2000) Land Cover Change in Australia. Results of

the Collaborative Bureau of Rural Sciences - State Agencies Project on

Remote Sensing of Land Cover Change. Bureau of Rural Sciences, Canberra.

Barton NH (2001) The evolutionary consequences of gene flow and local adaptation:

future approaches. In: Dispersal: Individual, Population and Community.

(eds. Clobert J, Danchin E, Dhondt A, and Nichols J), pp. 329-340 Oxford

University Press, Oxford.

Baverstock PR, Moritz C (1990) Sampling design. In: Molecular Systematics. (eds.

Hillis DM, and Moritz C), pp. 13-24 Sinauer Associates, Inc, Massachusettes.

Begon M, Harper JL, Townsend CR (1990) Ecology: Individuals, Populations and

Communities, Second. Blackwell Scientific Publications, Boston.

Belbin L (1994) PATN: Pattern analysis software Commonwealth Scientific and

Industrial Research Organisation.

Bilney RJ, Emison WB (1983) Breeding of the White Bellied Sea Eagle in the

Gippsland Lakes Region of Victoria, Australia. Australian Bird Watcher, Vol

10 (No.2), 61-68.

Birky CW, Fuerst P, Matuyama T (1989) Organelle gene diversity under migration,

mutation and drift: equilibrium expectations, approach to equilibrium,effects

of heteroplasmic cells and comparison to nuclear genes. Genetics, 121, 613-

627.

Birley A, Croft J (1986) Mitochondrial DNAs and phylogeopgraphic relationships.

In: DNA Systematics - Volume 1: Evolution. (eds. Dutta S), pp. Ch 5 CRC

Press, Inc. USA.

141 References

Blackburn TM, Gaston KJ (2002) Scale in macroecology. Global Ecology and

Biogeography, 11 (3), 185-189.

Blakers M, Davies SJJF, Reilly PN (1984) The Atlas of Australian Birds. Royal

Australian Ornithologists Union, Melbourne University Press., Victoria.

Bloom PH (1987) Capturing and handling raptors. In: Raptor Management

Techniques Manual. (eds. Given Pendleton BA, Millsap BA, Cline KW, and

Bird DM), pp. 99-123 National Wildlife Federation, Washington DC.

Bortolotti GR (1984) Sexual size dimorphism and age related size variation in Bald

Eagles. Journal of Wildlife Management, 48 (1), 72-81.

Brichetti P, Foschi UF, Boano G (2000) Does El Niño affect survival rate of

Mediterranean populations of Cory's Shearwater? Waterbirds, 23 (2), 147-

154.

Brown GG, Gadaleta G, Pepe G, Saccone C, Sbis E (1986) Structural conservation

and variation in the d-loop containing region of vertebrate mitochondrial

DNA. Journal of Molecular Biology, 192, 503-511.

Brown JH, Maurer BA (1989) Macroecology: the division of food and space among

species on continents. Science, 243, 1145-1150.

Brown JH, Mehlman DW, Stevens GC (1995) Spatial variation in abundance.

Ecology, 76 (7), 2028-2043.

Brown RD (1996) Attraction of Bald Eagles to habitats just below dams in Piedmont

North and South Carolina. In: Raptors in Human Landscapes: Adaptations to

Built and Cultivated Environments. (eds. Bird D, Varland D, and Negro J),

Academic Press Ltd., London.

142 References

Brown S, Schreier H, Thompson WA, Vertinsky I (1994) Linking Multiple Accounts

with GIS as a Decision Support System to Resolve Forestry/Wildlife

Conflicts. Journal of Environmental Management, 42, 349-364.

Brown WM, George MJR, Wilson AC (1979) Rapid evolution of animal

mitochondrial DNA. Proceedings of the National Academy of Science USA,

76 (4), 1967-1971.

Brown WM (1983) Evolution of animal mitochondrial DNA. In: Evolution of Genes

and Proteins. (eds. Nei M, and Koehn RK), Sinauer Assoc, Inc., Sunderland,

Massachusettes.

Bryan AL, Murphy TM, Bildstein KL, Brisbin IL, Mayer JJ (1996) Use of reservoirs

and other artificial impoundments by Bald Eagles in South Carolina. In:

Raptors in Human Landscapes: Adaptations to Built and Cultivated

Environments. (eds. Bird D, Varland D, and Negro J), pp. 285-298 Academic

Press Ltd., London.

Chakraborty R, Nei M (1977) Bottleneck effects on average heterozygosity and

genetic distance with the stepwise mutation model. Evolution, 31, 347-356.

Chamberlain D, Fuller R, Bunce R, Duckworth J, Shrubb M (2000) Changes in the

abundance of farmland birds in relation to the timing of agricultural

intensification in England and Wales. Journal of Applied Ecology, 37, 771-

788.

Chappell J (1983) A revised sea-level record for the last 300,000 years from Papua

New Guinea. Search, 14 (3-4), 99-101.

Clarke MF, Griffioen P, Loyn RH (1999) Australian Bird Count: Where do all the

bush birds go? Wingspan, 9 (4), I - XVI.

143 References

Clement M, Derington J, Posada D (2001) TCS: estimating gene geneologies.Version

1.13. Brigham Young University, Provo, UT 84602, US.

Clunie P (1994) Action Statement No. 60 - White-bellied Sea-Eagle. Department of

Conservation & Natural Resources.

Coffey M (1987) Sex determination of Bald Eagles by morphometric measurement.

U.S. Fish and Wildlife Service. Report No. 66-87.

Crandall KA, Templeton AR (1993) Empirical tests of some predictions from

coalescent theory and application to intraspecific phylogeny reconstruction.

Genetics, 134, 959-969.

Crochet P (2000) Genetic Structure of Avian Populations - Allozymes revisited.

Molecular Ecology, 9, 1463-1496.

Crow JF, Aoki K (1984) Group Selection for polygenic behavioural trait: a

differential proliferation model. Proceedings of the National Academy of

Science USA, 79, 2628-2631.

Delehanty DJ, Tybie RA, Ditsworth MJ, Hoelzer GA, Oring LW, Longmire JL

(1995) Genetic and morphological methods for gender identification of

mountain quail. Journal of Wildlife Management, 59 (4), 785-789.

Dennis TE, Lashmar AFC (1996) Distribution and abundance of White-bellied Sea-

Eagles in South Australia. Corella, 20 (3), 93-102.

Department of Environment and Heritage (2003) An Interim Biogeographic

Regionalisation for Australia Version 4.0 - Photographs. 2 September 2003

2003, .

144 References

Department of Natural Resources & Mines (2003) The Southern Oscillation Index. 20

March 2003 2003, Queensland Government,

lationIndex>.

Double M, Abbott C (1998) A report on the development of a DNA-based sexing

technique for the North Island kokako (Callaeas cinerae wilsoni). Division of

Botony and Zoology, Australian National University, ACT 0200, Australia.

Dowling TE, Moritz C, Palmer JD (1990) Nucleic acids II: restriction site analysis.

In: Molecular Systematics. (eds. Hillis DM, and Moritz C), pp. 250-317

Sinauer Assoc,Inc., Sunderland, Massachusettes.

Edwards SV (1993a) Mitochondrial gene geneology and gene flow among island and

mainland populations of a sedentary songbird, the grey-crowned babbler

(Pomatostomus temporalis). Evolution, 47 (4), 1118-1137.

Edwards SV (1993b) Long-distance gene flow in a cooperative breeder detected in

geneologies of mitochondrial DNA sequences. Proceedings of the Royal

Society of London. Series B, 252, 177-185.

Ellis S, Seal US (1995) Tools of The Trade To Aid Decision-Making For Species

Survival. Biodiversity and Conservation, 4, 553-572.

Emison WB, Bilney RJ (1982) Nesting habitat and nest site characteristics of the

White-Bellied Sea-Eagle in the Gippsland Lakes region of Victoria, Australia.

Raptor Research, 16 (2), 54-58.

Endler JA (1977) Geographic Variation, Speciation and Clines. Princeton University

Press, Princeton.

145 References

Endler JA (1986) Natural Selection in the Wild. Princeton University Press,

Princeton, New Jersey.

Ericson PG, Christidis L, Cooper A, Irestedt M, Jackson J, Johansson US, Norman JA

(2002) A Gondwanan origin of passerine birds supported by DNA sequences

of the endemic New Zealand wrens. Proceedings of the Royal Society of

London. Series B, 269, 235-241.

Erlich HA, Arnheim N (1992) Genetic analysis using the polymerase chain reaction.

Annual Review of Genetics, 26, 479-506.

ESRI (1996) ARCVIEW 3.2. Environmental Systems Research Institute, Inc.

Estoup A, Jarne P, Cornuet J (2002) Homology and mutation models at microsatellite

loci and their consequences for population genetic analysis. Molecular

Ecology, 11, 1591-1604.

Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred

from metric distances among DNA haplotypes: application to human

mitochondrial DNA restriction data. Genetics, 131, 479-491.

Excoffier L. (2004) Patterns of DNA sequence diversity and genetic structure after a

range expansion: lessons from the infinite-island model. Molecular Ecology,

13, 853-64.

Faanes CA, Bystrak D (1981) The role of observer bias in the North American

Breeding Bird Survey. Studies in Avian Biology, 6, 353-359.

Favaloro N (1944) The White-breasted Sea-Eagle along the Murray River. EMU, 43,

233-242.

Ferguson-Lees J, Christie DA (2001) Raptors of the World. Christopher Helm,

London.

146 References

Fitzpatrick BM, Dunk JR (1999) Ecogeographic variation in morphology of red-tailed

hawks in western North America. Journal of Raptor Research, 33 (4), 305-

312.

Flather CH, Sauer JR (1996) Using landscape ecology to test hypotheses about large-

scale abundance patterns in migratory birds. Ecology, 77 (1), 2835.

Flood N, Peacock A (1997) Twelve month Australian rainfall relative to historical

records. Queensland Centre for Climate Applications, Dept Natural Resources

- QLD.

Forero MG, Donazar JA, Blas J, Hiraldo F (1999) Causes and consequences of

territory change and breeding dispersal distance in the black kite. Ecology, 80

(4), 1298-1310.

Ford J (1987) Minor isolates and minor geographical barriers in avian speciation in

continental Australia. EMU, 87, 90-102.

Franklin DC (1999) Evidence of disarray among granivorous bird assemblages in the

savannas of northern Australia, a region of sparse human settlement.

Biological Conservation, 90, 53-68.

Futuyma DJ (1986) Evolutionary Biology, 2nd. Sinauer Associates Inc., Sunderland,

Massachusetts.

Gandon S, Michalakis Y (2001) Multiple causes of the evolution of dispersal. In:

Dispersal: Individual, Population and Community. (eds. Clobert J, Danchin E,

Dhondt A, and Nichols J), pp. 155-167 Oxford University Press, Oxford.

Garcelon DK, Martell MS, Regid PT, Buøen LC (1985) Morphometric, karyotypic,

and laproscopic techniques for determining sex in Bald Eagles. Journal of

Wildlife Management, 49 (3), 595-599.

147 References

Garnett S (2001) Atlas of Australian Birds Winners. Wingspan (December), 8-13.

Garnett S, Crowley G, Barrett G (2002a) Patterns and Trends in Australian Bird

Distributions and Abundance: Preliminary Analysis of Data from the New

Atlas of Australian Birds. Report to the National Land & Water Resources

Audit, Canberra.

Garnett S., Crowley G. & Barrett G. (2002b) Birds. In: Australian Terrestrial

Biodiversity Assessment 2002. (eds P. Sattler & C. Creighton) pp. 65-80.

National Land and Water Resources Audit, Commonwealth of Australia,

Canberra.

Gaston KJ, Lawton JH (1990) Effects of scale and habitat on the relationship between

regional distribution and local abundance. Oikos, 58, 329-335.

Gaston KJ, Blackburn TM, Greenwood JJ, Gregory RD, Quinn RM, Lawton JH

(2000) Abundance-occupancy relationships. Journal of Applied Ecology, 37

(Suppl. 1), 39-59.

Geoscience Australia (1990) Dams and Water Storages 1990 February 2002.

Commonwealth of Australia (Geoscience Australia - National Mapping

Division).

Glenn TC (1999) Extraction protocols. February 1999, Laboratory Of Molecular

Systematics, Smithsonian Institution, .

Glenn TC, Wolfgang S, Braun MJ (1999) Effects of a population bottleneck on

whooping crane mitochondrial DNA variation. Conservation Biology, 13 (5),

1097-1107.

148 References

Goodwin BJ, Fahrig L (1998) Spatial scaling and animal population dynamics. In:

Ecological Scale: theory and applications. (eds. Peterson DL, and Parker VT),

pp. 193-206 Columbia University Press, New York.

Gopurenko D, Hughes JM (2002) Regional patterns of genetic structure among

Australian populations of the mud crab, Scylla serrata (Crustacea: Decapoda):

evidence from mitochondrial DNA. Marine and Freshwater Research, 53,

849-857.

Grapputo A, Pilastro A, Marin G (1998) Genetic variation and bill size dimorphism in

a passerine bird, the reed bunting Embiriza schoeniclus. Molecular Ecology,

7, 1173-1182.

Greenberg R, Cordero PJ, Droege S, Fleisher RC (1998) Morphological adaptation

with no mitochondrial DNA differentiation in the Coastal Plain Swamp

Sparrow. The Auk, 115 (3), 706-712.

Greenwood AD, Pääbo S (1999) Nuclear insertion sequences of mitochondrial DNA

predominate in hair but not in blood of elephants. Molecular Ecology, 8, 133-

137.

Griffioen PA, Clarke MF (2002) Large-scale bird-movement patterns evident in

eastern Australian atlas data. EMU, 102, 99-125.

Griffiths R, Tiwari B (1995) Sex of the last wild Spix's macaw. Nature, 375, 452.

Griffiths R, Double MC, Orr K, Dawson RJG (1998) A DNA Test to Sex Most Birds.

Molecular Ecology, 7, 1071-1075.

Guo Q, Brown JH, Valone TJ (2000) Abundance and distribution of desert annuals:

are spatial and temporal patterns related? Journal of Ecology, 88, 551560.

149 References

Haig SM, Avise JC (1996) Avian conservation genetics. In: Conservation Genetics:

Case Histories from Nature. (eds. Avise JC, and Hamrick JL), pp. 160-189

Chapman & Hall, New York.

Hall T (1999) BIOEDIT: a user friendly biological sequence alignment editor and

analysis program for Windows 95/98/NT. Nucleic Acids Symposium Series,

41, 95-98.

Harrison RG (1989) Animal mitochondrial DNA as a genetic marker in population

and evolutionary biology. Trends in Ecology and Evolution, 4 (1), 6-11.

Hewitt GM (2000) The genetic legacy of quartenary ice ages. Nature, 405 (June),

907-913.

Hodson A (1992) Essential Genetics. Bloomsbury, London.

Hudson RR (1990) Gene geneologies and the coalescent process. In: Oxford surveys

in evolutionary biology. (eds. Futuyma DJ, and Antonovics J), pp. 1-44

Oxford University Press, Oxford.

Hughes JM, Baker AM, De Zylva G, Mather PB (2001) A phylogeographic analysis

of southern and eastern populations of the Australian magpie: evidence for

selection in maintenance of the distribution of two plumage morphs.

Biological Journal of the Linnean Society, 74 (1), 25-34.

Hunt WG (1998) Raptor floaters at Moffat's equilibrium. Oikos, 82 (1), 191-197.

Imre I, McLaughlin RL, Noakes DLG (2002) Phenotypic plasticity in brook charr:

changes in caudal fin induced by water flow. Journal of Fish Biology, 61,

1171-1181.

150 References

Innes JL (1998) Measuring environmental change. In: Ecological Scale: Theory and

Applications. (eds. Peterson DL, and Parker VT), pp. 429-457 Columbia

University Press, New York.

Jaksic FM, Lazo I (1999) Response of a bird assemblage in semi-arid Chile to the

1997-1998 El Niño. Wilson Bulletin, 111 (4), 527-535.

Jaksic FM (2001) Ecological effects of El Niño in terrestrial ecosystems of western

South America. Ecography, 24 (3), 241-250.

James FC (1983) Environmental component of morphological differentiation in birds.

Science, 221, 184 -186.

James FC, McCulloch CE, Wiedenfeld DA (1996) New approaches to the analysis of

population trends in land birds. Ecology, 77 (1), 13-27.

Jassby AD, Powell TM (1990) Detecting changes in ecological time series. Ecology,

71 (6), 2044-2052.

Karns DR, O'Bannon A, Voris HK, Weight LA (2000) Biogeographical implications

of mitochondrial DNA variation in the Bockadam snake (Cerberus rynchops,

Sepentes: Homalopsinae) in Southeast Asia. Journal of Biogeography, 27,

391-402.

Keast A (1981a) The evolutionary biogeography of Australian birds. In: Ecological

Biogeography of Australia. (eds. Keast A), pp. 1587-1635 Dr. W. Junk, The

Hague - Boston - London.

Keast A (1981b) Distributional patterns, regional biotas, and adaptations in the

Australian biota: a synthesis. In: Ecological Biogeography of Australia. (eds.

Keast A), pp. 1895-1997 Dr. W. Junk, The Hague - Boston - London.

151 References

Kenward RE, Walls SS, Hodder KH, Pahkala M, Freeman SN, Simpson VR (2000)

The prevelence of non-breeders in raptor populations: evidence from rings,

radio-tags and transect surveys. Oikos, 91, 271-279.

Kikkawa J, Monteith GB, Ingram G (1981) Cape York Peninsula: Major region of

faunal interchange. In: Ecological Biogeography of Australia. (eds. Keast A),

pp. 1697-1742 Dr. W. Junk, The Hague - Boston - London.

Kingsford R (2000) Ecological impacts of dams, water diversions and river

management on floodplain wetlands in Australia. Austral Ecology, 25, 109-

127.

Kingsford RT, Curtin AL, Porter J (1999) Water flows on Cooper Creek in arid

Australia determine 'boom' and 'bust' periods for waterbirds. Biological

Conservation, 88, 231-248.

Kirk DA, Hyslop C (1998) Population status and recent trends in canadian raptors: A

review. Biological Conservation, 83 (1), 91-118.

Knowles LL, Maddison WP (2002) Statistical phylogeography. Molecular Ecology,

11, 2623-2635.

Koehn J, Brumley A, Gehrke (2000) Managing the Impacts of Carp. Bureau of Rural

Sciences (Department of Agriculture and Fisheries - Australia), Canberra.

Kumar S, Tamura K, Nei M (1993) MEGA: Molecular evolutionary genetics analysis

2.0. Pennsylvania State University, University Park.

Lande R, Shannon S (1996) The role of genetic variation in adaptation and population

persistence in a changing environment. Evolution, 50 (1), 434-437.

152 References

Levin SA (1986) Pattern, scale and variability: an ecological perspective. In:

Community Ecology: Lecture Notes in Bioinformatics 77. (eds. Hastings A),

pp. 1-12 Springer-Verlag, Berlin.

Levin SA (1992) The problem of pattern and scale in ecology. Ecology, 73 (6), 1943-

1967.

Link WA, Sauer JR (1997) New approaches to the analysis of population trends in

land birds: comment. Ecology, 78 (8), 26322634.

Lunt DH, Whipple LE, Hyman BC (1998) Mitochondrial DNA variable number of

tandem repeats (VNTRs): utility and problems in molecular ecology.

Molecular Ecology, 7, 1441-1455.

Marchant S, Higgins P (1993) Handbook of Australian, New Zealand Birds and

Antarctic Birds. Oxford University Press, Melbourne.

Marzluff JM, Sallabanks R (1998a) Avian Conservation: Research and Management.

Island Press, USA.

Marzluff JM, Sallabanks R (1998b) Past approaches and future directions for avian

conservation biology. In: Avian Conservation: Research and Management.

(eds. Marzluff JM, and Sallabanks R), Island Press, Washington D.C.

Masterov VB (2000) Postnatal development of Stellar's Sea Eagles in sexing and

aging techniques. Pages 17-28 in Ueta M, and McGrady M, editors. First

symposium on Stellar's and White-tailed Sea Eagles in East Asia. Wild Bird

Society of Japan, Tokyo Japan.

Maurer BA, Villard M-A (1996) Special Feature - Continental scale ecology and

neotropical migratory birds: How to detect declines amid the noise. Ecology,

77 (1), 1-2.

153 References

McElroy D, Moran P, Bermingham E, Kornfield I (1991) REAP - The restriction

enzyme analysis package Version 4.0. Dept of Zoology, University of Maine.

McManus JW (1985) Marine speciation, tectonics and sea-level changes in Southeast

Asia. Pages 133-138. Proceedings of the Fifth International Coral Reef

Congress, Tahiti.

Mila B, Girman DJ, Kimura M, Smith TB (2000) Genetic evidence for the effect of

the postglacial population expansion of the phylogeography of a north

American Songbird. Proceedings of the Royal Society of London. Series B,

267, 1033-1040.

Mindell DP, editor. (1997) Avian Molecular Evolution and Systematics. Academic

Press, USA.

Mindell DP, Sorenson MD, Dimcheff DE (1998) Multiple Independent origins of

Mitochondrial Gene Order in Birds. Proceedings of the National Academy of

Science USA, 95, 10693-10697.

Mooney N, Brothers N (1986) Sea Eagles' greatest problem is nest disturbance, says

NPWS. FINTAS, 9, 39-41.

Mooney N (1998) Proximity limits of nesting White-Bellied Sea-Eagles to human

habitation and large bodies of water. Boobook, 18 (2), 32-33.

Moritz C, Dowling TE, Brown WM (1987) Evolution of animal mitochondrial DNA:

relevance for population biology and systematics. Annual Review in Ecology

and Systematics, 18, 269-292.

Moritz C (1994a) Applications of mitochondrial DNA analysis in conservation: a

critical review. Molecular Ecology, 3, 401-411.

154 References

Moritz C (1994b) Defining 'Evolutionarily Significant Units' for conservation. Trends

in Ecology and Evolution, 9 (10), 373-375.

Morrison JL, Maltbie M (1999) Methods for gender determination of Crested

Caracaras. Journal of Raptor Research, 33 (2), 128-133.

Nei M, Maruyama T, Chakraborty R (1975) The bottleneck effect and genetic

variability in populations. Evolution, 29(1), 1-10.

Nei M (1987) Molecular Evolutionary Systematics. Columbia University Press, New

York.

Neigel JE (2002) Is FST obsolete? Conservation Genetics, 3, 167-173.

Newton I (1991) Population limitation in birds of prey: a comparative approach. In:

Bird Population Studies: Relevance to Conservation and Management. (eds.

Perrins CM, Lebreton JD, and Hirons GJM), pp. 3-21 Oxford University

Press, Oxford.

Nichols RA, Hewitt GM (1994) The genetic consequences of long distance dispersal

during colonisation. Heredity, 72, 312-317.

Norris-Caneda KH, Elliott JD (1998) Sex identification in raptors using PCR. Journal

of Raptor Research, 32 (4), 278-280.

O'Brien SJ, Martenson JS, Miththapala S, Janczewski D, Pecon-Slattery J, Johnson

W, Gilbert DA, Roelke M, Packer C, Bush M, Wildt DE (1996) Conservation

genetics of the felidae. In: Conservation Genetics: Case Histories from

Nature. (eds. Avise JC, and Hamrick JL), pp. 50-74 Chapman & Hall, New

York.

Olendorff RR (1972) Weighing and measuring raptors. Raptor Research, 6 (2), 53-56.

155 References

Olsen PD (1995) Australian Birds of Prey: The Biology and Ecology of Raptors.

University of New South Wales Press, Sydney.

Olsen PD (1998) Australia's pest animals: New solutions to old problems. Kangaroo

Press, Roseville.

Page RDM, Holmes EC (1998) Molecular Evolution: A Phylogenetic Approach.

Blackwell Science, Oxford.

Palumbi S, Martin A, Romano S, Owen McMillan W, Stice L (1991) The simple

fool's guide to PCR. Version 2.0. Department of Zoology and Kewalo Marine

Laboratory, University of Hawaii.

Peacock MM, Ray C (2001) Dispersal in pikas (Ochotona princeps): combining

genetic and demographic approaches to reveal spatial and temporal patterns.

In: Dispersal. (eds. Clobert J, Danchin E, Dhondt A, and Nichols J), pp. 43-56

Oxford University Press Inc., New York.

Peters DE (1979) Some evidence for a decline in population status of the Regent

Honeyeater. Australian Bird Watcher, 8 (4), 117-123.

Posada D, Crandall KA, Templeton AR (2000) GeoDis: a program for the cladistic

nested analysis of the geographical distribution of genetic haplotypes.

Molecular Ecology, 9, 487-488.

Quinn GP, Keough MJ (2002) Experimental design and data analysis for biologists.

Cambridge University Press, Cambridge, United Kingdom.

Quinn TW (1992) The genetic legacy of mother goose: phylogeographic patterns of

lesser snow goose Chen caerulescens caerulescens maternal lineages.

Molecular Ecology, 1, 105-117.

156 References

Rand DM (1996) Neutrality tests of molecular markers and the connection between

DNA polymorphism, demography, and conservation biology. Conservation

Biology, 10 (2), 665-671.

Raphael MG, Evans Mack D, Cooper BA (2002) Landscape-scale relationships

between abundance of marbled murrelets and distribution of nesting habitat.

The Condor, 104, 331-342.

Ray N, Currat M, Excoffier L (2003) Intra-deme molecular diversity in spatially

expanding populations. Molecular Biology and Evolution, 20 (1), 76-86.

Reid JRW (2000) Threatened and Declining Birds in the New South Wales Sheep-

Wheat Belt: II. Landscape Relationships - Modelling Bird Atlas Data Against

Vegetation Cover. Consultancy report to NSW National Parks and Wildlife

Service. CSIRO Sustainable Ecosytems.

Ridpath MG, Brooker MG (1986) The breeding of the Wedge-tailed Eagle Aquila

audax in relation to its food supply in arid Western Australia. Ibis, 128, 177-

194.

Robards F (1967) Capture, Handling and Banding Bald Eagles. US Dept. Int., Bur.

Sport Fish. and Wildl.

Robertson A, Simmons RE, Jarvis AM, Brown CJ (1995) Can bird atlas data be used

to estimate population size? A case study using Namibian endemics.

Biological Conservation, 71, 87-95.

Robertson G (1987) Effect of drought on a breeding population of Wedge-tailed

Eagles Aquila audax. EMU, 87, 220-223.

157 References

Rogers AR, Harpending H (1992) Population growth makes waves in the distribution

of pairwise genetic differences. Molecular Biology and Evolution, 9 (3), 552-

569.

Roshier DA, Whetton PH, Allan RJ, Robertson AI (2001) Distribution and

persistence of temporary wetland habitats in arid Australia in relation to

climate. Austral Ecology, 26 (4), 371-384.

Rozas J, Rozas R (1999) DnaSP Version 3: an integrated program for molecular

population genetics and molecular evolutionary analysis. Bioinformatics, 15,

174-175.

SAS (1996) SAS System Version 6.12 for Windows 6.12. SAS Institute Inc.

Schneider S, Excoffier L (1999) Estimation of demographic parameters from the

distribution of pairwise differences when the mutation rates vary among sites:

Application to human mitochondrial DNA. Genetics, 152, 1079-1089.

Schneider S, Roessli D, Excoffier L (2000) ARLEQUIN Version 2.0. Genetics and

Biometry Laboratory, Dept of Anthropology and Ecology, University of

Geneva.

Schoener TW (1987) The geographical distribution of rarity. Oecologia, 74, 161-173.

Seibold I, Helbig AJ (1996) Phylogenetic relationships of the sea eagles (genus

Haliaeetus): reconstructions based on morphology, allozymes and

mitochondrial DNA sequences. Journal of Zoological Systematics and

Evolutionary Research, 34, 103-112.

Seutin G, Ratcliffe LM, Boag PT (1995) Mitochondrial DNA homogeneity in the

phenotypically diverse redpoll finch complex (Aves: Carduelinae: Carduelis

Flammea-Hornemanni). Evolution, 49 (5), 962-973.

158 References

Shephard JM, Hughes JM, Zalucki MP (2002) Genetic differentiation between

Australian and North American populations of the monarch butterfly Danaus

plexippus (L.) (Lepidoptera: Nymphalidae): an exploration using allozyme

electrophoresis. Biological Journal of the Linnean Society, 75, 437-452.

Siriwardina GM, Baillie SR, Buckland ST, Fewster RM, Marchant JH, Wilson JD

(1998) Trends in the abundance of farmland birds: a quantitative comparison

of smoothed Common Birds Census indices. Journal of Applied Ecology, 35,

24-43.

Slatkin M (1987) Gene Flow and the Geographic Structure of Natural Populations.

Science, 236, 787-792.

Slatkin M, Maddison WP (1989) A cladistic measure of gene flow inferred from the

phylogenies of alleles. Genetics, 123, 603-613.

Smith DG, Wiemeyer SN (1992) Determining sex of Eastern Screech-Owls using

discriminant function analysis. Raptor Research, 26 (1), 24-26.

Smith-Gill SJ (1983) Developmental plasticity: developmental conversion versus

phenotypic modulation. American Zoologist, 23, 47-55.

Stockwell CA, Mulvey M, Vinyard GL (1996) Translocations and the Preservation of

Allelic Diversity. Conservation Biology, 10(4), 1133-1141.

Stokes T (1996) Helicopter effects upon nesting White-bellied Sea-Eagles and upon

smaller birds at an isolated protected location (Eshelby Island, Great Barrier

Reef, Australia). Corella, 20 (1), 25-28.

Storfer A (1996) Quantitative Genetics: a promising approach for the assessment of

genetic variation in endangered species. Trends in Ecology and Evolution, 11

(8), 343-348.

159 References

Strimmer K, von Haeseler A (1996) Quartet puzzling: a quartet maximum likelihood

method for reconstructing tree topologies. Molecular Biology and Evolution,

13, 964-969.

Swain DP, Foote CJ (1999) Stocks and chameleons: the use of phenotypic variation

in stock identification. Fisheries Research, 43 (1-3), 113-128.

Tabachnick BG, Fidell LS (2001) Using Multivariate Statistics, Fouth Edition. Allyn

& Bacon, Boston.

Tajima F (1989) Statistical method for testing the neutral mutation hypothesis by

DNA polymorphism. Genetics, 123, 585-595.

Tamura K, Nei M (1993) Estimation of the number of nucleotide substitutions in the

control region of the mitochondrial DNA in humans and chimpanzees.

Molecular Biology and Evolution, 10, 512-526.

Taylor BL, Chivers SJ, Sexton S, Dizon AE (2000) Evaluating dispersal estimates

using mtDNA data: Comparing analytical and simulation approaches.

Conservation Biology, 14 (5), 1287-1297.

Templeton AR, Boerwinkle E, Sing CF (1987) A cladistic analysis of phenotypic

associations with haplotypes inferred from restriction endonuclease mapping.

I. Basic theory and an analysis of alcohol dehydrogenase activity in

drosophila. Genetics, 117, 343-351.

Templeton AR, Sing CF (1993) A cladistic analysis of phenotypic associations with

haplotypes inferred from restriction endonuclease mapping. IV. Nested

analyses with cladogram uncertainty and recombination. Genetics, 134, 659-

669.

160 References

Templeton AR, Routman E, Phillips CA (1995) Separating population structure from

population history: A cladistic analysis of the geographical distribution of

mitochondrial DNA haplotypes in the tiger salamander, Ambystoma tigrinum.

Genetics, 140, 767-782.

Templeton AR (1998) Nested clade analyses of phylogeographic data: testing

hypotheses about gene flow and population history. Molecular Ecology, 7,

381-397.

Templeton A. R. (2004) Statistical phylogeography: methods of evaluating and

minimizing inference errors. Molecular Ecology, 13, 789-809.

Thackway R, Cresswell ID (1995) An Interim Biogeographic Regionalisation of

Australia: A Framework for Setting Priorities in the National Reserves System

Cooperative Program. Version 4.0. Australian Nature Conservation Agency,

Reserve Systems Unit.

Thomas L (1996) Monitoring Long-term population change: why are there so many

analysis methods ? Ecology, 77 (1), 49-58.

Thurstans S (1998) Modelling the nesting habitat of the White-Bellied Sea-Eagle,

Haliaeetus leucogaster. Bachelor of Science with Honours Thesis. University

of Tasmania, Hobart.

Tubaro PL, Segura ET (1995) Geographic, ecological and subspecific variation in the

song of the rufous-browed peppershrike (Cyclarhis gujanensis). Condor, 97,

792-803.

Turner MG, Gardner RH, O'Neill RV (2001) Landscape Ecology in Theory and

Practice: Pattern and Process. Springer-Verlag, New York.

161 References

Vallianatos M, Lougheed SC, Boag PT (2002) Conservation genetics of the

loggerhead shrike (Lanius ludovivianus) in central and eastern North America.

Conservation Genetics, 3, 1-13.

Veltman EJ, Carrick R (1990) Male biased dispersal in Australian Magpies. Animal

Behaviour, 40 (1), 190-192.

Weir BS, Cockerham C (1984) Estimating F-Statistics for the analysis of population

structure. Evolution, 38(6), 1358-1370.

Wenink PW, Baker AJ, Rosner H-U, Tilanus MG (1996) Global mitochondrial DNA

phylogeography of holarctic breeding dunlins (Calidris alpina). Evolution, 50

(1), 318-331.

Williams A (1997) Report on the Monitoring of the White-Bellied Sea-Eagle

Haliaeetus leucogaster in the Gippsland Region -1996. Report for the

Department of Natural Resources and Environment. March 19 1997.

Williams A (1998a) Survey of the White-Bellied Sea-Eagle Haliaeetus leucogaster in

the Gippsland Region 1997. Report by the Department of Natural Resources

and the Environment - Bairnsdale. Feb 17 1998.

Williams A (1998b) Monitoring of the White-Bellied Sea Eagle Haliaeetus

leucogaster in the Gippsland Region 1998. East Gippsland Flora and Fauna

Report No.2. Flora and Fauna Group, Department of Natural Resources and

Environment, Bairnsdale.

Wright S (1951) The genetical structure of populations. Annals of Eugenics, 15, 323-

354.

Zavalaga CB, Frere E, Gandini P (2002) Status of the red-legged Cormorant in Peru:

What factors affect distribution and numbers? Waterbirds, 25 (1), 8-15.

162 References

Zink R, Remsen J (1986) Evolutionary processes and patterns of geographic variation

in birds. In: Current Ornithology. (eds. Johnston R), pp. 1-69 Plenun Press,

N.Y.

Zink R (1994) The geography of mitochondrial DNA variation, population structure,

and species limits in the fox sparrow (Passerella iliaca). Evolution, 48, 96-

111.

Zink RM (1997) Phylogeographic studies of North American birds. In: Avian

molecular evolution and systematics. (eds. Mindell DP), pp. 301-324

Academic Press, San Diego.

Zink RM, Barrowclough GF, Atwood JL, Blackwell RRC (2000) Genetics,

taxonomy, and conservation of the threatened California Gnatcatcher.

Conservation Biology, 14 (5), 1394-1405.

163 Appendix

Appendices

164 Appendix 1

Appendix 1 – Source of Museum specimens.

Source State No. Location Type Age @ 2000 Putative Sex Genetic sex Age Museum Spec No. Hobart Museum TAS Tas 28 Port Arthur Skin 18 Female F Juvenile Hobart Museum TAS Tas 29 Satellite Is.- Bruny Is. Carcass 4 Juvenile Hobart Museum TAS Tas 30 Bust-Me-Gall Hill Carcass 4 Female Adult Hobart Museum TAS Tas 31 Cockle Bay Skin 18 Female Juvenile Hobart Museum TAS Tas 32 Marion Bay Skin 14 Female Adult CO1350 Hobart Museum TAS Tas 33 Rokeby Skin Immature 8805/B2558 Hobart Museum TAS Tas 34 Maria Island Skin 21 Male Immature B3523/CO326. Hobart Museum TAS Tas 35 Swansea Skin 83 Male Immature B727/2075 Launceston Museum TAS Tas 17 Pegara SF Res - King Is. Skin 21 Female F Adult Launceston Museum TAS Tas 18 Unknown - north ? Carcass M Adult Launceston Museum TAS Tas 19 Armstrongs Lane - Cressy Carcass 3 Female Immature Launceston Museum TAS Tas 20 Lakes District Carcass 5 Male M Juvenile Launceston Museum TAS Tas 21 Unknown - north ? Skin 43 FAIL Adult Launceston Museum TAS Tas 22 Tamar River Skin 23 Male Juvenile Launceston Museum TAS Tas 23 Rocky Cape Skin 14 Female F Juvenile Launceston Museum TAS Tas 24 Winnaleah Skin 39 Female FAIL Adult Launceston Museum TAS Tas 25 St Patricks River Skin 95 Male FAIL Adult Launceston Museum TAS Tas 26 Moulting Lagoon Skin 34 Male F Juvenile Launceston Museum TAS Tas 27 Unknown - north ? Skin 93 Male M Immature NT - Museum NT NT 8 Kapalga Skin 16 Female F 0046/0498 Qld Museum TAS Tas 39 Launceston Skin 47 Male Adult 4917 Qld Museum QLD Qld 5 Queensland Skin Female Young Adult 30213 Qld Museum QLD Qld 6 Brisbane Skin 54 Young Adult 4368 Qld Museum QLD Qld 7 Brisbane - Jimboomba Skin 93 Female F Adult 28865 Qld Museum QLD Qld 8 Cardwell - NE Qld Skin 2 Female Juvenile 17740 Qld Museum QLD Qld 9 Unknown - presumably Qld Skin Immature 10230 Qld Museum QLD Qld 10 Townsville Skin Male Adult 28775 Qld Museum QLD Qld 11 Mouth of Brisbane River Skin 46 Female Juvenile 5433 Qld Museum QLD Qld 12 Moreton Island Skin 2 Female F Adult 27089 Qld Museum QLD Qld 13 Cooktown - Annan River Skin 40 Female Immature Qld Museum PNG PNG 1 PNG - Conflict Group Skin 112 Adult 19374 Qld Museum PNG PNG 2 PNG - East Coast (used Lae) Skin 112 Adult 19281 Qld Museum PNG PNG 3 PNG - East Coast (used Lae) Skin 112 Adult 19282 SA Museum SA SA18 Port Lincoln NP Skin M B48773 SA Museum SA SA19 Antechamber Bay (Kangaroo Is.) Skin B45988 SA Museum SA SA20 Wisanger Park (Kangaroo Is.) Skin M B46632

165 Appendix 1

Source State No. Location Type Age @ 2000 Putative Sex Genetic sex Age Museum Spec No. SA Museum SA SA21 Djim Djim Rock (Northern Territory) Skin B27758 SA Museum SA SA 22 Ballast Head (Kangaroo Is.) Skin B46337 Vic Museum TAS Tas 40 Bass Strait - Flinders Is. Skin 87 Female Juvenile R.6060 Vic Museum TAS Tas 41 Maria Island Skin 86 Female Immature R.6116 Vic Museum QLD Qld 1 North Queensland Skin 89 Adult B16501 Vic Museum QLD Qld 2 North Qld - Barrier Reef Skin Male Immature B.16503 Vic Museum QLD Qld 17 Rockhampton Skin Juvenile R.1391 Vic Museum VIC Vic 1 Tambo Upper Skin 8 Male M Young Adult B.19513 Vic Museum VIC Vic 2 Western Port Skin Young Adult R.1390 Vic Museum VIC Vic 3 Queenscliff Skin 138 Male Adult 12615 Vic Museum VIC Vic 4 Victoria Skin 86 Female Immature R.6059 Vic Museum VIC Vic 5 Gippsl Lakes - Rotamah Is. Skin 122 Male Juvenile R.1393 Vic Museum VIC Vic 6 Blond Bay Game Reserve Skin 10 Female F Adult B.19540 Vic Museum VIC Vic 23 Morwell Tissue 2 M Vic Museum WA WA 1 Etordis Ck - Northern WA Skin 85 Female Adult 6562 Vic Museum WA WA 2 Napier, Broome Bay Skin 90 Female Adult 5294 Vic Museum NSW NSW 1 Mathoura, Crompt's Lane Skin 8 Male M 1445 Vic Museum NSW NSW 2 Mooney Ck near Coff's Harbour Carcass 8 Female F Immature MV2350 Vic Museum NT NT 1 Mouth of McCarthur River, Gulf of Carpentaria Skin 89 Juvenile B16502 Vic Museum NT NT 2 Port Essington Skin 140 Male Adult B16504 WA Museum WA WA 9 Torbay Inlet Skin 91 Male Adult 11388 WA Museum WA WA 10 Ida Bay, Cape Arid Skin 94 Male Immature 8870 WA Museum WA WA 11 Barrow Island Skin 99 Female Immature 3219 WA Museum WA WA 12 Abrolhos Is Skin ? Female Adult 375 WA Museum WA WA 13 North Twin Peak Is Skin 94 Female Adult 8359 WA Museum WA WA 14 Pt Cloats Skin 98 Female Immature 5092 WA Museum WA WA 15 Vasse Skin 72 Female Immature A3259 WA Museum WA WA 16 Dampier Salt Skin 12 Female Immature A22002 WA Museum WA WA 17 Dorrie Island Skin 90 Male Adult 10516 WA Museum WA WA 18 Bernier Is Skin 90 Male Adult 10517 WA Museum WA WA 19 Derby Skin 102 Male Adult 273 WA Museum WA WA 20 Abrolhos Is Skin 102 Male Immature 371 WA Museum WA WA 21 Dirk Hartog Is Skin 84 Male Immature A1162 WA Museum WA WA 22 Dorre Is Skin 90 Chick 10519

166