Extinction Risk and Population

Declines in Amphibians

Jon Bielby

Imperial College London,

Division of Biology,

A thesis submitted for the degree of Doctor of Philosophy at

Imperial College London

– May 2008

1

Thesis abstract

This thesis is about understanding the processes that explain the patterns of extinction risk and declines that we see in amphibians, how we can use that understanding to set conservation priorities, and how we can convert those priorities into practical, hands-on research and management. In particular, I focus on the threat posed by the emerging infectious disease, chytridiomycosis, which is caused by the chytrid fungus, Batrachochytrium dendrobatidis ( Bd ).

Amphibians display a non-random pattern of extinction risk, both taxonomically and geographically. In chapter two I investigate the mechanism behind the observed taxonomic selectivity and find that it is due to species biology rather than heterogeneity in either threat intensity or conservation knowledge.

In chapter three I determine which biological and environmental traits are important in rendering a species susceptible to declines, focussing on susceptibility to Bd . I found that restricted range, high elevation species with an aquatic life-stage are more likely to have suffered a decline. Using these traits, I predict species and locations that may be susceptible in the future, and which should therefore be a high priority for amphibian research and conservation.

2 The use of predictive models to set conservation priorities relies on the accuracy of the modelling technique used. In chapter four I compare the predictive performance of both phylogenetic and nonphylogenetic regression and decision trees, and assess the suitability of each technique. Although phylogenetic regression provided the most predictive models of extinction risk and decline, decision trees could be a useful tool in future studies.

Once high priority species or locations have been identified as susceptible to

Bd , what should be the next step? In chapter five I identify highly susceptible species that have suffered an increase in extinction risk as a result of threat processes other than chytridiomycosis, and investigate the potential role of the disease in their decline. I find that the historical range of one of the four species investigated does harbour Bd at a relatively high prevalence, which suggests that structured surveillance of Bd in this location, and further investigation of the role of chytridiomycosis in the species’ decline would be warranted.

In chapter six I describe such a structured, opportunistic screening for Bd on the island of Sardinia. I assess the threat that chytridiomycosis may pose to the threatened and endemic species of Sardinia and discuss future research directions and conservation needs of the island’s amphibians.

3 Declaration

All of the work presented in this thesis is mine, with the following acknowledgements:

Chapter two

This chapter was published as a senior authored paper in Animal

Conservation (Bielby, Cunningham, & Purvis, 2006, Animal Conservation,

9,135-143). The main idea behind the paper was my own, although Andy

Purvis provided guidance on how to conduct the analyses. Olivia Curno,

Sanjay Molur, Shawn McCracken of www.tadpole.org , Ronald Nussbaum of the University of Michigan, Rohan Pethiyagoda of the Wildlife Heritage Trust of Sri Lanka, Mark-Oliver Rödel, staff at Jatun Sacha research centre, Bruce

Young of NatureServe helped with gathering site species check-lists. All aspects of data collection and collation, analysis, interpretation and writing are my own.

Chapter three

This chapter was published as a senior authored paper in Conservation

Letters (Bielby, Cooper, Cunningham, Garner & Purvis, Conservation Letters, in press ). Natalie Cooper helped collect the data used in the analysis, and

Andy Purvis provided guidance on the statistical techniques used. Tim

Coulson provided advice on the statistical methods used. Emmanuel Paradis provided advice on the use of compare.gee function. S. Blair Hedges,

Roberto Ibanez, Andres Merino-Viteri, Eugenia del Pino, Santiago Ron, and

Tej Kumar Shrestha all helped in compiling data on the number of eggs per

4 clutch. The idea behind the chapter and all aspects of data analysis, interpretation and writing are my own.

Chapter four

The ideas behind the analyses in this chapter were developed by me with guidance from Andy Purvis. Natalie Cooper and Marcel Cardillo provided some of the phylogenetic comparative analyses presented in the chapter, and are acknowledged accordingly. The amphibian dataset underpinning the analyses were collected by me in collaboration with Natalie Cooper. The mammalian data analysed in the chapter was collected as part of the

PanTheria project. All aspects of data analysis, interpretation and writing are my own.

Chapter five

The main idea behind this chapter is my own. Patricia Burrowes, Rafael

Joglar, and Abel Vale helped with the field work conducted in Puerto Rico.

Tissue samples from amphibians were provided by Stefan Lötters, Trent

Garner, and Thomas Doherty-Bone. Barry Clarke allowed access to the NHM collection. Frankie Clare helped to process the laboratory samples. All aspects of data interpretation and writing are my own.

Chapter six

A grant awarded to Trent Garner, Stefano Bovero from the People’s Trust for

Endangered Species provided funding for this chapter. The Sardinian NGO

Zirrichiltaggi and their associates provided invaluable logistical help, sausage,

5 cheese and wine that proved essential for the fieldwork conducted in Sardinia.

The Forestry workers and managers of Sardinia also provided invaluable support during my time on the island. Frankie Clare helped with the processing of the laboratory samples. All aspects of analysis, interpretation and writing are my own.

6 Acknowledgements

Thanks to NERC for the funding that made my doctoral research possible.

Also thanks to a grant from PTES which funded the work on Sardinia presented in this thesis.

Thanks also to Simon Stuart and Janice Chanson provided help with use of the Global Amphibian Assessment database.

To start with, big thanks to my supervision team, Andrew, Trent, and Andy.

Before I started, someone pointed out that by the end I’d almost certainly hate my supervisors at least a little bit. Well, I can honestly say that has not happened as yet. Maybe after the viva. Thanks in particular to Andy, who

I’ve now known for a long time, and who has given me real guidance, direction and confidence over the past few years.

A famous writer once said: “All first drafts are shit. They are, however, necessary shit.” So a big thank you to all of the people who in the last three and a half years have read through the necessary and highlighted my

“opportunistic” use of commas, my generally misuse of semi-colons, my chess-like use of paragraphs (“hhhmmm, I’ve not used the knight for a while, now seems like a good time”), my over-use of hyphens, and my generally poor grammar. These people include the many members of the Purvis lab

7 (too many to mention/remember) since 2004, and also Amanda Duffus, and

Andrew King.

There are, of course, periods of time during a PhD when all sense of reality goes out of the window. In those times, my mates were there to see me through. Thanks to all of those at Silwood, IoZ, and the Real World who have slapped me back down to earth every so often. Charlie, Shaun, Jimmy, Matty and Neil have always given me a sense of perspective when needed; thanks lads. Also, special thanks go to: Guy Cowlishaw who has helped me hold it together on a couple of occasions when I was about to come apart at the seams; 1001 shukrans to Trent and Dina who were basically always there for me, and fed me fish and wine when I dropped by like The Littlest Hobo; Ben and Janna who always had a pie on hand in case of an emergency; and a special firm-handshake to my office-mates for much of my PhD, Kingy and

Amber who’ve……well, let’s not get all emotional. But thanks.

Finally thanks to all of my family who’ve provided support, interest and encouragement. In particular, thanks to Auntie Jan and Uncle Ron for those days as Little Switzerland catching newts and eating Drewery’s hot-cakes.

That’s pretty much where it all started, so thank you for being around and being interested and transferring some of your enthusiasm and energy to me.

And finally thanks to Mum, Dad and Ali, because all this PhD really represents is me doing things that I like and care about and enjoy, and that’s what you’ve always encouraged me to do. Thank you.

8 Table of Contents

Thesis abstract...... 2 Declaration ...... 4 Acknowledgements ...... 7 Table of Contents...... 9 List of Tables...... 12 List of Figures ...... 14 List of Appendices...... 15 List of Acronyms and Abbreviations………………………………………….16

Chapter 1: General Introduction ...... 18 Chapter 2: Taxonomic Selectivity in Amphibians: Ignorance, Geography or Biology?...... 28 Abstract...... 28 Introduction...... 29 Methods...... 32 Data ...... 32 Analyses ...... 35 Results ...... 37 Discussion ...... 38 Tables ...... 44 Appendix ...... 53 Chapter 3: Predicting susceptibility to future declines in the world’s frogs...... 60 Abstract...... 60 Introduction...... 61 Methods...... 65 Data ...... 65 Analyses ...... 67 Results ...... 71 Discussion ...... 72 Tables ...... 79 Figure ...... 83

9 Appendices ...... 84 Chapter 4: Modelling extinction risk in multispecies data sets: phylogenetically independent contrasts vs. decision trees...... 113 Abstract...... 113 Introduction...... 114 Methods...... 119 Data ...... 119 Analyses ...... 121 Predictive performance ...... 124 Results ...... 126 Discussion ...... 129 Tables ...... 134 Figure ...... 137 Appendices ...... 138 Chapter 5: Ground-testing a predictive model of Bd -susceptibility with the aim of directing further research...... 161 Abstract...... 161 Introduction...... 162 Methods...... 165 Target species ...... 165 Collection of samples ...... 168 Type of tissue samples taken ...... 168 Molecular screening of samples ...... 171 Results ...... 171 Discussion ...... 173 Tables ...... 181 Appendix ...... 185 Chapter 6: Distribution and epidemiology of Batrachochytrium dendrobatidis on the Mediterranean island of Sardinia...... 186 Abstract...... 186 Introduction...... 187 Methods...... 191 Study system ...... 191

10 Sample sizes ...... 192 Target Species ...... 192 Collection and field hygiene ...... 194 Laboratory techniques ...... 195 Results ...... 195 Discussion ...... 198 Table ...... 205 Figures ...... 206 Chapter 7: General Conclusions ...... 212 References...... 221

11 List of Tables

Table 2.1: Results of chi-square test for random distribution of threatened species at the global level, and subsequent analysis of distribution of threatened species between amphibian families after the omission of DD species…………………………………………………………………….. 44

Table 2.2: Results of country level Fisher’s exact test and subsequent analysis of distribution of threatened species between amphibian families after the omission of DD species………………………………………………46

Table 2.3: Results of site level Fisher’s exact test analysis, and subsequent analysis of distribution of threatened species between amphibian families after the omission of DD species………………………………………………50

Table 2.4: Combined p-values obtained from Fisher’s method, illustrating the tendency for families to be over- or under-threatened within different countries……………………………………………………………………52

Table 3.1: Results of univariate and multipredictor analyses of RD species vs. threatened species………………………………………………………..79

Table 3.2: Results of univariate and multipredictor analysis of enigmatic RD species vs. threatened species…………………………………………. 80

Table 3.3: Results of univariate and multipredictor analysis of Bd + RD species vs. all other Bd + species………………………………………………….81

Table 3.4: Results of univariate and multipredictor analysis of Bd + RD species vs Bd - RD species………………………………………………………. 82

12 Table 4.1: Details of the comparisons made and the optimal tree size for each.……………………………………………………………………….134 Table 4.2: Summary of technique performance in terms of consistency, precision, and presence of a phylogenetic signal…………………….. 135

Table 4.3: Interaction terms added to the PCM regression models based on the corresponding DT………………………………………………….....136

Table 5.1: Summary of Bd surveillance results for Bufo brauni ……. 181

Table 5.2: Summary of Bd surveillance results for Bufo lemur at El Tallonal, Puerto Rico………………………………………………………………...182

Table 5.3: Summary of Bd surveillance results for Hyperolius cystocandicans in central Kenya………………………………………………………….. 183

Table 5.4: Summary of Bd surveillance results for Xenopus longipes at Lake Oku, Cameroon…………………………………………………………... 184

Table 6.1: Details of populations of Sardinian amphibians sampled in the 2007 field-season………………………………………………………… 205

13 List of Figures

Fig. 3.1: Global distribution of anuran species with a predicted probability of Bd -related decline =1…………………………………………………… 83

Fig. 4.1: Significant differences in MSE among techniques for each comparison……………………………………………….. ……………….137

Fig. 6.1: Locations of all Sardinian populations sampled in the 2007 field season……………………………………………………………………...206

Fig. 6.2: Locations of sampled populations in the southern Sette Fratelli mountain region of Sardinia…………………………………………….. 207

Fig. 6.3: One of 10 specimens of Euproctus platycephalus with damaged digits……………………………………………………………….. ………208

Fig. 6.4: Locations of sampled populations in the northern Limbara mountain region……………………………………………………………………… 209

Fig. 6.5: Locations of infected sites in the Limbara mountain region. …………………………………………………………………………...... 210

Fig. 6.6: Locations of sampled populations in the Gennargentu mountain region……………………………………………………………………… 211

14 List of Appendices

Appendix 2.1: Site species check-lists that were not suitable for analysis. ………………………………………………………………………………53 Appendix 3.1: References used in anuran data collection……………84

Appendix 3.2: List of Bd+ species………………………………………92

Appendix 3.3: Results of ANOVAs used to determine whether there were significant differences between anuran families in the explanatory variables used in the analyses……………………………………………….…….. 112

Appendix 4.1: Frog regression models…………….…………...... 138

Appendix 4.2: Mammalian regression models……………………….. 143

Appendix 4.3: Optimal decision tree models…………….…………….148

Appendix 4.4: Correlations coefficients of fitted values obtained from three different modelling techniques for ten different comparisons…………158

Appendix 4.5: Mean squared error of extinction risk models obtained from different modelling techniques for ten different clades………………..159

Appendix 4.6: Results of analysis of model residuals for phylogenetic signal. ………………………………………………………………………………160 Appendix 5.1. Sample size required for detecting disease where the probability of finding at least one case in the sample is 5%...... 185

15 List of Acronyms and Abbreviations

ACAP – Amphibian Conservation Action Plan AET – Actual Evapotranspiration a.s.l. – Above Sea Level Bd – Batrachochytrium dendrobatidis CR – Critically Endangered DD – Data Deficient DNA – Dexoyribonucleic acid DT – Decision Tree EN – Endangered EPH – Endemic Pathogen Hypothesis ETI – External Threat Index GAA – Global Amphibian Assessment GEE – Generalised Estimating Equations GE – Genomic Equivalent GIS – Geographic Information System HPD – Human Population Density IPC – Internal Positive Control IUCN – Internationl Union for Conservation of Nature MAM – Minimum Adequate Model MSE – Mean Squared Error MU – Management Units NGO – Non-governmental Organisation NPH – Novel Pathogen Hypothesis NPP – Net Primary Productivity NT – Near Threatened PCM – Phylogenetic Comparative Methods PTES – People’s Trust for Endangered Species qPCR – Quantitative Polymerase Chain Reaction RD – Rapid Decline

16 TIPS – Non-phylogenetic Regression analysis using species as independent datapoints UVB – Ultraviolet B VU – Vulnerable

17 Chapter 1: General Introduction

The extinction crisis

The Earth’s biodiversity is facing a crisis, with current rates of species extinction estimated to be much higher than the historical rate in several taxonomic groups (Mace et al. 2005). Although biodiversity is defined and measured on scales ranging from genes to ecosystems, conservation attention is frequently focussed at the species level. The species level focus is largely due to the practical advantages of using this unit of biodiversity: most species are relatively easy to recognise (Baillie et al. 2004; but see Isaac et al. 2004; Mace 2004); they are well suited to describing patterns in biodiversity (Brooks et al. 2004); and they may be used to infer changes in the state of biodiversity at other scales (e.g. Butchart et al. 2004).

With species being the focal point of so much conservation attention it is necessary to have a robust metric with which we can assess their status.

The IUCN Red List (IUCN 2001; IUCN 2008) is the most widely recognised and accepted method available for measuring a species’ risk of extinction (De

Grammont and Cuaron 2006; Rodrigues et al. 2006). The Red List implements objective criteria, based on information on population size, rate of decline and size of geographic distribution, to assign species to categories of risk of extinction ranging from “Least Concern” through to “Extinct” (IUCN

2001). The Red List now contains assessments for over 41500 species, including almost all birds, mammals, amphibians, conifers and cycads (IUCN

2008).

18

Amphibians are the most recent large clade to have been systematically assessed and introduced into the Red List (IUCN et al. 2004) and, relative to other fully assessed groups, they are not faring well. 32% of amphibian species are threatened with extinction compared to 23% of mammals and

12% of birds (IUCN 2004). Additionally, 23% of amphibian species were classified as Data Deficient (DD), many of which are likely to be threatened

(Akçakaya et al. 2000). Not only are a high proportion of amphibian species threatened, but in recent years there has been a decline of amphibian species on all continents on which they live (ACAP 2005). It appears that not all amphibian species are equally affected by extinction risk or population declines: there is disparity in the proportion of both threatened and declining species among clades (Baillie et al. 2004; Stuart et al. 2004), and certain biological traits have been linked to species’ risk of local decline (Hero et al.

2005; Lips et al. 2003) and extinction (Cooper et al. 2008; Murray and Hose

2005). In gaining a better understanding of the mechanisms responsible for differences in extinction risk and decline among amphibian lineages, we may be able to mitigate the impacts of the active threat processes more effectively.

Several threat processes, such as habitat loss, overexploitation (Stuart et al.

2004), and introduced predators (Kats and Ferrer 2003), have been linked to amphibian declines. Additionally, in recent years, declines and even extinctions of species as a result of “enigmatic” threat processes have been described in a several locations. A variety of mechanisms, including pollution

(Davidson 2004), UVB irradiation (Kiesecker et al. 2001), and climate change

19 (Whitfield et al. 2007) have been linked to these declines in amphibians.

However, much recent attention has focussed on the role of the emerging infectious disease, chytridiomycosis (Berger et al. 1998), caused by the fungal pathogen Batrachochytrium dendrobatidis (hereafter Bd ; Longcore et al.

1999).

Bd , a member of the phylum Chytridiomycota, was described as a species as recently as 1999 (Longcore et al.). Generally, chytrid fungi are widespread and common in nature, playing an important role in degrading substances such as chitin, keratin, and plant detritus in the environment (Berger et al.

1998). Although chytrid fungi are known to parasitize a range of hosts, including plants, fungi and invertebrates, Bd is the first reported case of a chytrid parasitizing a vertebrate host (Berger et al. 1998; Longcore et al.

1999). Bd is aquatic, with water being essential for the survival and propagation of the pathogen. It infects keratinised areas of the amphibian host by means of a motile, waterborne zoospore. Infection typically occurs in either the mouthparts of the larvae, the only keratinised body region in this lifestage, or the epidermis of fully developed individuals (Berger et al. 1998;

Marantelli et al. 2004). Upon infection the zoospores develop into sporangia which in turn produce and discharge more zoospores (Berger et al. 2005a).

Infection does not appear to be host-specific; Bd has been recorded as having infected over 450 species on five continents

(http://www.parcplace.org/bdmap2008update.html). Further, the response to infection may vary both inter- and intraspecifically (Rachowicz et al. 2006), with some species seemingly acting as asymptomatic vectors (Daszak et al.

20 2004), while others may suffer mortalities (Berger et al. 1998), and population declines (La Marca et al. 2005). After such population declines, some populations or species may persist (Retallick et al. 2004), whereas others may experience local or even species extinction as a result of infection (Schloegel et al. 2006).

The amount of research attention that the pathogen has attracted is perhaps a result of the potential rapidity and severity of its effects (ACAP 2005; Lips et al. 2006), and the fact that many of the declines in which Bd has been implicated have occurred in relatively undisturbed habitat in protected areas

(Laurance et al. 1996; Lips 1998; Pounds and Crump 1994). Combined, these characteristics of Bd -related decline suggest that traditional conservation efforts may not be sufficient to reduce the impact of the disease: more focussed, novel management strategies may be required. Such strategies include the establishment of ex situ populations (Mendelson et al.

2006a), restriction of the multi-million dollar trade in amphibians (Fisher and

Garner 2007), and raising awareness of the possible modes of pathogen dissemination (Johnson and Speare 2003; Johnson and Speare 2005).

However, instigating such management and policy is likely to be species- and location-specific, labour intensive and expensive. Determining which species are the most susceptible to the negative effects of Bd in the future, and may therefore be a high priority for conservation actions, is therefore an important and urgent objective.

21 Given the scale and rate at which we are losing amphibian populations and species, our ability to understand the patterns of decline and extinction that we see and to tackle the underpinning causal mechanisms has assumed a great deal of importance. In this thesis I aim to contribute to our knowledge of the patterns of extinction risk that we observe, and the mechanisms underpinning those patterns; I also aim to outline how this knowledge may inform conservation efforts

Overview of thesis aims

In this thesis I aim to gain a better understanding of the processes underpinning the observed taxonomic pattern of extinction risk, identify which species are likely to be the most susceptible to future increases in extinction risk, and evaluate how our predictions may vary according to the particular modelling technique used. However, the knowledge gained is largely academic unless it can be used to inform decisions on management or policy, and it is not always immediately obvious how the link between academic research and applied conservation is best fostered. I therefore use this thesis, in part, to illustrate how the output of predictive models may be used to direct further research with a view to informing decisions on the direction of future policy, management and research. In order to do so, I focus upon amphibians and aim to answer four questions. First, why are some clades more highly- threatened than others? Second, what type of species are the most susceptible to future increases in extinction risk? Third, which technique is the most accurate in predicting species’ susceptibility to future elevation in extinction risk? Lastly, how can we use our knowledge of species

22 susceptibility to direct future research efforts and the resulting on the ground management and policy?

Chapter organisation and objectives

My thesis is wide in scope, geographically, taxonomically, and analytically. It begins at the global-scale, observing, describing and explaining global patterns in taxonomic non-randomness in extinction risk at higher taxonomic levels. Remaining at the global-scale, my thesis then moves to a lower taxonomic level and focuses on testing which biological and environmental traits are associated with the observed patterns of species decline. Using the insight gained from the pattern analysis my thesis then makes the step from pattern analysis to prediction, by predicting which species will be most susceptible to decline in the future. In my final two empirical chapters, I then illustrate how predictive models may be tested using directed field observations on particular species, before finally moving into an example of more detailed population monitoring and disease surveillance that may be used to inform and direct conservation action.

In chapter two, I address the subject of taxonomic selectivity (non- randomness) in extinction risk. Specifically, I use permutation tests on different data treatments at several spatial scales to test for the presence of selectivity, and to determine which of three suggested mechanisms contribute to its presence. Heterogeneity in conservation knowledge, threat intensity and distribution of biological traits among clades may all contribute to non-random distribution of extinction risk. Since understanding which mechanisms are

23 responsible for the pattern may be useful in conservation planning, these hypotheses each need to be tested.

After demonstrating the taxonomic non-randomness of extinction risk, I next ask which type of species are most likely to be susceptible to future declines.

Several studies conducted in a diverse range of taxonomic groups have shown associations between amphibian species’ risk of extinction or decline and information on their biology, environment, and the intensity of threat to which they are exposed (see Fisher and Owens 2004; Lips et al. 2003; Sodhi et al. 2008; Williams and Hero 1998). In chapter three, I ask which biological and environmental characteristics affect amphibian species’ susceptibility to increases in Red List status. In the 2004 Global Amphibian Assessment

(hereafter GAA; IUCN et al. 2004), 435 species, termed Rapid Decline (RD) species (Stuart et al. 2004), were highlighted as having suffered an increase in Red List status. By looking for consistent associations between the RD status of species and biological and environmental traits, I aim to determine which species are susceptible to decline in the future. Focussing on those biological traits associated with decline as a result of infection by the pathogen

Bd , I then predict which species outside of the model-building dataset are most likely to be susceptible to Bd related decline in the future. Although correlation does not mean causation, and the results of such tests must therefore be interpreted with care, the predictions highlight susceptible species and identify locations with a high number of those species with a view to informing applied conservation efforts.

24 If models are to be useful in predicting a species’ level of risk or likelihood of decline, we must be sure that their predictions are accurate and reliable. The majority of studies seeking to model extinction risk have used linear model techniques, often incorporating some kind of phylogenetic comparative method (PCM). The widespread use of PCM in models of extinction risk stems from the need to account for phylogenetic non-independence of species level data, and the resulting increased rates of type I errors

(Felsenstein 1985). Recently, however, there has been some discussion as to whether phylogenetic methods are required in studies of extinction risk

(Putland 2005); also, an alternative method to linear models, decision trees, has been advocated as an ideal way to model a species’ level of extinction risk (Jones et al. 2006; Sullivan et al. 2006). The advantages of decision trees include their ease of use and ability to highlight some of the more complex relationships among variables (Crawley 2002; De'ath and Fabricius

2000; Sullivan et al. 2006). In chapter four, I compare and contrast the predictive performance of different modelling techniques, and make recommendations for future efforts to model extinction risk. Specifically, I assess the predictive accuracy of three techniques (PCM regression, TIPS regression, and Decision Trees) in modelling Rapid Decline and Red List status of frogs, and extinction risk in five mammalian clades. Based on the results of the comparisons of predictive performance, I make recommendations for future efforts to model extinction risk and decline.

The output of a predictive model will typically be a list of susceptible species, or a geographic representation of the intensity of threat or a measure of

25 species richness (e.g. Cardillo et al. 2006). In chapters five and six I outline how such an output may be used to direct research, and how in turn more focussed research may be used to inform policy and management of a particular conservation issue. In chapter five I highlight four species from the model of Bd –susceptibility produced in chapter three, each of which have suffered either a decline or a mass mortality. I then use tissue samples from those species to investigate the presence of Bd and hence the possibility that the pathogen could have been implicated in those declines. Although the presence of Bd does not prove that the pathogen is the cause of species’ decline (e.g. Daszak et al. 2005), it does indicate that further investigation is worthwhile.

In chapter six, I describe a detailed surveillance scheme for Bd to illustrate how knowledge of the distribution and epidemiology of the pathogen may inform policy and management. Bd was first described on the island of

Sardinia in 2006, when infected specimens of the Sardinian brook salamander

(Euproctus platycephalus ) in the south of the island displayed signs of chytridiomycosis (Bovero et al. in press ). I describe a systematic screening of amphibian populations on Sardinia aimed at improving our knowledge of the distribution and epidemiology of Bd on the island. As effective management for preventing the introduction and dissemination of Bd may rely on knowledge of the origin, distribution and mode of spread of the pathogen

(Rachowicz et al. 2005), such in-depth studies may be invaluable for informing the decision making process and implementing the measures necessary for limiting the impacts of Bd within a region.

26

The final chapter summarises and synthesises my main findings. I argue that, despite some idiosyncrasies that will occur as a result of the scale of the analyses and factors not included in these predictive models, large-scale models can be informative in directing future research and conservation management and policy. Further, I make recommendations on improving the performance and application of predictive models with a view to providing focussed direction for the conservation of amphibian species.

27 Chapter 2: Taxonomic Selectivity in Amphibians: Ignorance,

Geography or Biology?

The manuscript presented in this chapter is published as:

Bielby, J., Cunningham, A.A., & Purvis, A. (2006), Animal Conservation,

9,135-143

Abstract

Many taxa, including amphibians, have been shown to have a taxonomically non-random distribution of threatened species. There are three possible non- exclusive reasons for this selectivity: non-random knowledge of species conservation status, clades endemic to different regions experiencing different intensities of threatening process, and the effects of clade-specific biological attributes on the susceptibility of species to those processes. This chapter tests the sufficiency of the first two explanations using extinction risk evaluations from the 2004 Global Amphibian Assessment. My results indicate that they cannot alone account for the degree of non-randomness in extinction risk among amphibian families. The overall distribution of threatened amphibians remained taxonomically non-random when species of unknown conservation status were omitted, and significant selectivity was detected not only at a global geographic scale, but also in country- and site-specific data sets. Furthermore, the same families tend to be over- or under-threatened within different countries. Together, these results suggest that biological

28 differences among amphibian families play an important role in determining species’ susceptibility to anthropogenic threatening processes.

Introduction

Information available in IUCN Red Lists, such as overall level of threat, prevalence and intensity of threatening processes, and geographic and taxonomic distribution of threatened species, can have important implications for conservation policy and practice. The recent GAA (IUCN et al. 2004) provided the first comprehensive Red List evaluation of amphibian species.

An analysis of the assessments (Baillie et al. 2004) found extinction risk to be non-randomly distributed among amphibian families; some families have significantly more threatened species (i.e. species categorised as being

Vulnerable, Endangered, or Critically Endangered) than expected and others have fewer, a pattern previously reported for mammals and birds (Bennett and

Owens 1997; Purvis et al. 2000a; Russell et al. 1998). Likewise, threatened and Near Threatened amphibian species that are thought to be rapidly declining were not evenly distributed between families (Stuart et al. 2004).

Such a non-random pattern has been termed “taxonomic selectivity”, and has been observed in species extinctions as well as extinction risk (McKinney

1997).

What causes such selectivity? Explanations so far have centred around the biological characteristics of lineages that make them more or less susceptible to the mechanisms causing elevated risk of extinction (Bennett and Owens

29 1997; Cardillo et al. 2004; Jones et al. 2003; McKinney 1997; Owens and

Bennett 2000; Purvis et al. 2000b). Here I consider two other possible explanations.

The first possibility is that non-random patterns in risk are simply a consequence of taxonomically non-random ignorance about the conservation status of species. Taxonomic biases exist in both conservation research

(Bonnett et al. 2002) and practice (Seddon et al. 2005), with some clades receiving more resource per species than others. The presence of such biases suggests that the level of knowledge of species conservation status could also be non-random, with species within some orders or families being poorly understood relative to others. In the context of Red Lists, species for which there is insufficient knowledge to make a full conservation assessment are categorised as being “Data Deficient” (hereafter DD) (IUCN 2001). Some amphibian families, such as Bufonidae, have a relatively small proportion of

DD species (59 of 461, 12% of species), whereas others such as Caeciliidae have a much higher proportion (66 of 109, 60% of species) (IUCN et al. 2004).

In the analyses of distribution of threatened and rapidly declining species in amphibians (Baillie et al. 2004; Stuart et al. 2004), DD species were treated as though non-threatened. In previously-analysed taxa, few species are listed as DD (birds 0.8%, mammals 5.3%) so the way in which they are treated is unlikely to affect the results. By contrast, 22.5% of amphibian species are classified as DD, so their treatment could have a large impact on the results.

If DD species are treated as either all threatened or non-threatened, because of the uneven distribution of DD species across amphibian families, the

30 analysis could give a false impression of threat prevalence within at least some families, and hence of heterogeneity of threat prevalence among families.

The second possible cause of apparent nonrandomness is that where, rather than how, a species lives determines its risk of extinction (Russell et al. 1998).

Different locations have different intensities of threatening processes such as habitat degradation, over-exploitation, invasive species and infectious disease

(Baillie et al. 2004; Stuart et al. 2004). The prevalence of extinction risk might be higher in more heavily affected locations, where all species may be exposed to higher levels of threat (but see Balmford 1996; Kerr and Currie

1995; McKinney 2001). If, additionally, particular clades are more or less restricted to locations with a high threat intensity, we might expect them to have a higher than average threat prevalence regardless of their biology, leading to a taxonomically non-random pattern of global extinction risk. This mechanism is of particular relevance in amphibians as they display more endemicity than mammals or birds: 70% of amphibian species have restricted geographic ranges (restricted range being a range of less than 50,000km 2:

BirdLife International 2004a) compared to 25% of birds (IUCN et al. 2004), with many families also having a relatively small distribution. For example, the family Astylosternidae has a high prevalence of threat (21 of 29 species;

72%), and is restricted to West Africa, predominantly in habitats that are heavily affected by agriculture and deforestation (Baillie et al. 2004). Do such families have a high prevalence of threat solely because of their geographic location? Although this mechanism has been in the literature for some years

31 (Russell et al. 1998), it has not to my knowledge been explicitly addressed previously, perhaps because relatively few bird and mammal families are narrow-range endemics. The high levels of endemism in amphibians make it a potentially powerful cause of selectivity in extinction risk within the clade.

In this paper I aim to determine whether the global pattern of non-random extinction risk persists after the omission of DD species. I also assess whether selectivity is independent of geographic differences by analysing extinction risk at finer geographic scales, where the intensity of threatening processes is more homogenous than at the global level. If the level of extinction risk were purely due to the different intensities of threatening processes among geographic locations, at a fine-scale each family should display a similar proportion of threatened species. Persistence of a taxonomically non-random pattern of extinction risk at small geographic scales would therefore suggest that the non-randomness is not entirely due to geography. Such a result would represent strong evidence that biological differences are at least partly responsible for the observed selectivity, indicating the potential for insightful analyses of correlates of extinction risk.

Methods

Data

The 2004 IUCN Red List assessments of the world’s amphibian species were obtained from the 2004 GAA database ( http://www.globalamphibians.org ).

The database used Frost’s (2004) “Amphibian Species of the World” as its

32 taxonomic basis. This taxonomy describes three amphibian orders which together contain 48 families and 5473 species. For my analyses I used the same version of the database as Baillie et al. (2004) and Stuart et al. (2004) to allow direct comparison of results. This was a pre-launch version of the 2004

IUCN Red List assessments which differed from the finalised database for a few species endemic to one country (Brazil): these differences were unlikely to greatly affect the results. In order to remove the effects of taxonomically non-random ignorance of conservation status, I omitted DD species from the analysis to produce a global dataset of 4453 amphibian species, 1891 of them threatened or extinct.

To assess whether selectivity in extinction risk results solely from geographic differences in threatening processes, data sets were also constructed at two finer spatial scales: countries and sites. Sites were defined as areas or locations that were small enough to be conserved in their entirety (Alliance for

Zero Extinction 2005; BirdLife International 2004b). Country and site data sets were subsets of the global data set, containing only those species listed as occurring in the relevant country or site. Site species lists were obtained from peer-reviewed literature, research station check-lists, websites, field herpetologists, and GAA participants. Suitable data sets were selected using the following criteria:

- Data sets must contain at least 30 species and ideally at least 10 threatened species; this threshold was chosen to balance the conflicting aims of obtaining a reasonable number of data sets and obtaining enough statistical power to

33 identify deviations from a taxonomically random distribution of threatened species.

- At least two amphibian families should be represented in the subset, one of which should have enough species to offer sufficient statistical power to detect families that were significantly over or under-threatened relative to the subset as a whole.

- To be treated as independent, data sets at a given scale should have little overlap in species lists and not be geographically close.

- Selected countries should have a relatively small geographic area, to maximise the likelihood of homogeneous threatening processes within each

(largest analysed = 587,040km 2).

- Countries were chosen if they contained a suitable site data set. One of my sites is in Australia. However, at over an order of magnitude larger than the other countries analysed, Australia itself was too large to be meaningfully included. I therefore included the state (Queensland) as a further country- level data set, despite its large geographic area (1.7 x 10 6km 2).

The above criteria were met by eleven countries (Cameroon, Costa Rica,

Cuba, Ecuador, Japan, Madagascar, Malaysia, Queensland, Spain, Sri Lanka,

Uruguay) and six sites (Mount Kinabalu, Malaysia (Malkmus et al. 2002),

Nkongsamba (Amiet 1975), Korup National Park (Lawson 1993), and Mount

Nlonako (Herrmann et al. 2005), Cameroon, La Selva, Costa Rica (Donnelly

1994), Australian Wet Tropics Australia (Williams and Hero 1998)). Three of these sites were from the same country (Cameroon), and contained a large degree of overlap in constituent species. In order to maintain independence

34 of datasets only the results from the largest of these datasets, that of Mont

Nlonako (Herrmann et al. 2005), were included in this paper. However, results of the analyses of Nkongsamba and Korup (after amendments suggested by J.-L.Amiet (IUCN et al. 2004)) were consistent with those obtained from analysis of Mont Nlonako. Many other site lists were found

(see Appendix 2.1) but contained either too few species or too few threatened species for meaningful hypothesis tests.

Analyses

The aim of the first analysis was to test for deviations, at each spatial scale, from the null expectation that threatened species are distributed randomly among families. Within each data set, numbers of threatened and non- threatened species in each family were tabulated. For the global data set, these numbers were tested for variation in threat levels among families using a chi-square test; at the country and site level Fisher’s Exact Test was more suitable given the sample sizes involved (Crawley 2002). Fisher’s Exact

Tests were therefore conducted on each dataset within each of these two spatial scales. Within a spatial scale the p-values from contingency table analysis of each dataset were combined using Fisher’s method (Sokal and

Rolf 1995), treating the datasets as independent, to provide an overall test of significance of extinction risk selectivity at that spatial scale.

In the presence of taxonomically non-random extinction risk, further analyses were conducted with the aim of determining which families deviated from the expected level of threat. Significant deviation could only be detected for

35 families containing more than a threshold number of species, which depends on the proportion of threatened species in the data set being analysed. For each dataset, binomial-tests were used to determine the minimum family size necessary to detect a significant deviation from the proportion of threatened species. Families with fewer than the threshold number of species were not tested in the subsequent analysis. For remaining families, a null frequency distribution for the number of threatened species was obtained by 10000 unconstrained randomisations using a computer program written in the statistical language R (Ihaka and Gentleman 1996): Red List data were shuffled across all species in the data set, and the number of species in the family that were assigned a threatened category (CR, EN or VU) was counted.

The actual number of threatened species was then compared with this null frequency distribution, and the null hypothesis rejected if it lay in the 2.5% at either tail of the null distribution.

In order to determine whether families were consistent in the direction of deviation from expected levels of threat, for families present in three or more countries, the p-values obtained for each family being over-threatened (Table

2.2, column 3) and each family being under-threatened (Table 2.2, column 4) were combined across countries using Fisher’s Method (Sokal and Rolf 1995).

A combined probability of <0.05 suggested that the deviation was significantly different from random across countries and that the direction of deviation was consistent at that spatial scale. Due to the shortage of suitable datasets, consistency in direction of deviation could only be analysed at the country level.

36 Results

The global data set showed a significant deviation from a random distribution of threatened species among amphibian families (Table 2.1).

Fisher’s Exact Tests indicated that 7/11 countries analysed and 2/4 sites showed selectivity of extinction risk (Tables 2.2 and 2.3). Combining the p- values within each spatial scale using Fisher’s method indicated selectivity of risk at both the country ( χ 2 = 159.8, d.f. = 22, p=<0.0001) and site ( χ2 = 23.2, d.f. = 8, p=0.0014) level.

Of the 48 families in the global data set, 29 had enough species to depart significantly from the overall average extinction risk prevalence.

Randomisation tests showed seven of these to be over-threatened and ten to be under-threatened (Table 2.1).

Of the 17 families that were over or under-threatened at the global level, 11 differed significantly from random within at least one country (Table 2.2). Five families were analysable within at least three countries; each of these was significantly either over-threatened or under-threatened, compared with co- occurring families, when p-values from the separate countries were combined

(Table 2.4), indicating that the direction in which families deviated from random was consistent among countries. Three families deviated significantly from expected threat-levels within at least one site (Table 2.3). No families were significantly non-random exclusively at a lower spatial-scale (e.g. if a

37 family was non-random at a site level, it was also non-random at the country and global level).

Discussion

The persistence of selectivity in extinction risk after the omission of DD species (Table 2.1) strongly supports the conclusion of Baillie et al . (2004) and shows that their results were not purely an artefact of taxonomic bias in knowledge of conservation status. Further, the persistent non-randomness at both the country and site level (Tables 2.2 and 2.3) suggests that geographic differences in the intensity of threatening processes were not solely responsible for the selectivity observed at the global level. I discuss each of these findings in turn. There are also a number of amphibian families with insufficient species to statistically show a deviation, whose member species are either all threatened (Rheobatrachidae, Rhinodermatidae, Sooglossidae,

Nasikabatrachidae and Leiopelmatidae), or all non-threatened (Allophrynidae,

Amphiumidae, Ascaphidae, Dicamptodontidae, Pelodytidae, and

Rhinophrynidae). Regardless of their lack of significance statistically, these families represent evolutionarily distinct lineages, and their status supports the overall finding of taxonomic selectivity.

Exclusion of DD species did not remove all taxonomic selectivity, but did affect which families were found to be significantly over- or under-threatened.

The results for five families showed a qualitative difference between my study and that of Baillie et al . (2004). Centrolenidae (51 threatened species out of

138) did not depart from the overall mean in Baillie et al. ’s (2004) analysis, but

38 are overthreatened once the 49 DD species are excluded. Similarly,

Ichthyophiidae had been previously found to have unusually low prevalence of extinction risk, with just 2 of 39 species listed as threatened; however, 32 of the species are DD, with just 7 species of known conservation status, and the family does not depart significantly from the overall proportion of threatened species in my analysis. The results for families with a very large proportion of

DD species such as Ichthyophiidae, are unlikely to be accurate reflections of their true conservation status, but they illustrate how the treatment of DD species can affect the perceived level of threat within a clade. The changed results in the other three families are not due to removal of their DD species, but to the effect of the removal of all DD species on the mean prevalence of threat. Scaphiopodidae has no DD species, and Limnodynastidae only two, but my analyses find them to be under-threatened because the overall threat prevalence has risen from 26% in Baillie et al .’s (2004) analyses to 33% here.

Also, Ranidae (630 species, 112 DD, 165 threatened) is under-threatened in my analysis for this reason.

Most earlier analyses of selectivity have treated DD species as unthreatened, whereas I removed them. Other possible data treatments, such as treating all

DD species as threatened, or randomly allocating DD species to Red List categories to match the proportions within non-DD species, would also yield different results. Which of these treatments reflects the actual situation in amphibians most accurately? Classifying all DD species as non-threatened is highly conservative and, while it may be valid for policy-aimed documents

(e.g., Baillie et al. 2004), does not give a true reflection of the 1294 DD

39 species’ actual conservation status. Amphibians are much more poorly studied than mammals or birds (22.5% DD c.f. 5.3% mammals and 0.8% birds), with many species being poorly described, having a high level of taxonomic uncertainty, or being known purely from the holotype. For many such amphibian species the categorisation as DD means that we simply do not know where on the spectrum of extinction risk these species fit. Whereas in other taxa that have been fully assessed for the Red List (e.g. mammals and birds) it is likely that the majority of DD species are actually threatened

(Akçakaya et al. 2000), for amphibians, the treatment of all DD species as being threatened is unlikely to be an accurate reflection of their true conservation status. Random allocation of DD species to Red List categories in accordance with the proportions in non-DD species would maintain the size of the dataset being analysed. However, there is no evidence to suggest that the proportion of DD species that would fall into each category matches that of non-DD species, and the uneven taxonomic and geographic distribution of DD species (IUCN et al. 2004) argues to the contrary. Exclusion of DD species means that results and conclusions are derived solely from species for which there was sufficient information to make conservation assessments, which seems preferable to attempting to make generalisations about the conservation status of the less well-known species.

The presence of selectivity in amphibian extinction risk at a range of spatial scales suggests that geographic differences in threat intensity are not solely responsible for the non-random pattern at the global level: once much of the heterogeneity of threat intensity is accounted for, a taxonomically non-random

40 distribution of threatened species remains (it is not possible to account for all of the heterogeneity in threat, because it is likely to show some patchiness even within sites). Many amphibian families, however, have a relatively narrow distribution, making it possible that in some cases the distribution of the species within the country or site analysed was the total global distribution.

If this were the case any deviation from the null distribution at the global level would be a reflection of the situation at the finer-scale. However, of the 11 families that differed from random at the country level, none were endemic to that particular country.

The non-random distribution of risk showed consistency among countries, with the same families repeatedly being over-threatened (Leptodactylidae,

Bufonidae and Rhacophoridae) or under-threatened (Hylidae, Microhylidae and Ranidae) (Table 2.4). Ideally the consistency of direction of deviation would have been analysed at the site level. Unfortunately, I was unable to find enough site-level datasets. Many other site species lists were considered

(see Appendix 2.1), but they either did not have enough species in total, or contained too few threatened species for the analysis to have any power.

This paucity of data may reflect the low amphibian diversity, particularly of threatened species, in sites that have been surveyed; it is also possible that surveys tend to catalogue only the more widespread, common species, with highly-threatened narrow range endemics rarely being catalogued without specific efforts to do so.

41 The consistency of amphibian families in the direction of deviation from random expectations supports the remaining mechanism for taxonomic selectivity in extinction risk; namely that biological differences among taxa may interact with threatening processes to determine their extinction risk prevalence (Purvis et al. 2005b). Over-threatened families such as Bufonidae and Leptodactylidae contain genera with a relatively high proportion of threatened species such as Atelopus and Eleutherodactylus respectively.

There is evidence that these two genera are susceptible to interactions between chytridiomycosis and climate change (Burrowes et al. 2004; La

Marca et al. 2005; Pounds and Crump 1994; Pounds et al. 1999), which is likely to be due to certain aspects of their biology(Lips et al. 2003). Not all families containing such over or under threatened sub-clades will be highlighted by my analysis. The relatively high level of biological variation within amphibian families means that family-level analyses will not necessarily highlight which families contain species with biological characteristics that infer high or low susceptibility to threatening processes. It is therefore important that correlative analyses are conducted at the generic, or preferably the species level, in order to determine which biological traits underpin this mechanism. In other vertebrate groups, traits such as large body mass

(Bennett and Owens 1997), low population density (Cardillo et al. 2004), and low reproductive output (Bennett and Owens 1997; Cardillo et al. 2004) being significantly correlated with high extinction risk. Previous correlative work on amphibians has suggested that species with traits such as low fecundity, high degree of habitat specialisation (Williams and Hero 1998), large body size and highly aquatic lifestyle (Lips et al. 2003) are more likely to suffer population

42 declines. The biology of species is also likely to contribute to their ability to recover after such declines, and therefore might play a role in determining their risk of extinction. The consistency of deviation from randomness for amphibian families at all spatial scales analysed suggests either a common global driver of increased extinction risk, or that the same biological traits infer susceptibility to multiple drivers of amphibian decline (e.g. chytridiomycosis

(Berger et al. 1998; Daszak et al. 2003; Lips 1999; Lips et al. 2003), climate change (Pounds 2001; Pounds and Crump 1994; Pounds et al. 1999), habitat loss(Stuart et al. 2004), and interactions amongst these causes (Burrowes et al. 2004; Kiesecker et al. 2001)).

Distinguishing between trait selectivity and geographic selectivity is difficult if the intensity of threatening processes and the taxonomic mix of species vary greatly among regions (Russell et al. 1998), as is the case in amphibians, which show a higher degree of endemicity than mammals or birds. Testing for geographic selectivity is more complex than testing for trait selectivity, because a taxonomically random distribution of extinction risk would not be an appropriate null hypothesis. Although the level of threatening processes at a given geographic location will probably affect the prevalence of threatened species (Kerr and Currie 1995; McKinney 2001), the consistent tendency for particular amphibian families to be over-or under-threatened within countries, and the selectivity in extinction risk at all spatial scales, strongly suggest that biological differences among taxa play a role in determining their extinction risk prevalence, independently of geographical differences in intensity of threatening processes.

43 Tables

Table 2.1: Results of chi-square test for random distribution of threatened species at the global level, and subsequent analysis of distribution of threatened species between amphibian families after the omission of DD species. In all tables families are placed in descending order of the proportion of threatened species. N/A denotes that the family did not have enough species to reject the null hypothesis.

(Chi-square test, χ2 = 374.3, d.f.= N/A, simulated p=0.0001)

>Expected < Expected Family Threatened Total threat-level p- threat-level species species value p-value

Astylosternidae 21 28 0.000 1.000

Hynobiidae 27 40 0.001 1.000

Leptodactylidae 529 906 0.000 1.000

Plethodontidae 169 294 0.000 1.000

Centrolenidae 51 89 0.003 0.999

Rhacophoridae 113 201 0.000 1.000

Bufonidae 210 402 0.000 1.000

Bombinatoridae 5 10 0.428 0.794

Ambystomatidae 13 27 0.342 0.787

Dendrobatidae 65 136 0.115 0.914

Megophryidae 43 95 0.338 0.739

Arthroleptidae 13 34 0.747 0.372

44 Salamandridae 21 61 0.919 0.127

Discoglossidae 4 12 0.828 0.370

Heleophrynidae 2 6 0.808 N/A

Proteidae 2 6 0.800 N/A

Petropedetidae 24 73 0.963 0.061

Ranidae 165 518 1.000 0.000

Hylidae 211 686 1.000 0.000

Ichthyophiidae 2 7 0.874 0.363

Mantellidae 35 123 0.999 0.001

Microhylidae 71 261 1.000 0.000

Hyperoliidae 49 195 1.000 0.000

Limnodynastidae 10 48 1.000 0.001

Myobatrachidae 13 65 1.000 0.000

Hemisotidae 1 5 0.937 N/A

Pipidae 3 23 1.000 0.003

Caeciliidae 2 43 1.000 0.000

Scaphiopodidae 0 7 1.000 0.022

45 Table 2.2: Results of country level Fisher’s exact test and subsequent analysis of distribution of threatened species between amphibian families after the omission of DD species.

>Expected < Expected Species Species Family threat-level threat-level threatened Number p-value p-value

Cameroon Astylosternidae 20 26 0 1.00

(Fisher’s Bufonidae 10 22 0.05 0.98

Exact Test, Arthroleptidae 6 17 0.32 0.84

p=7.44x10 -8) Petropedetidae 4 18 0.81 0.39

Ranidae 4 27 0.98 0.07

Pipidae 1 9 0.95 N/A

Hyperoliidae 5 53 1.00 0.0003

Caeciliidae 0 2 N/A N/A

Hemisotidae 0 2 N/A N/A

Microhylidae 0 1 N/A N/A

Rhacophoridae 0 1 N/A N/A

Costa Rica Plethodontidae 20 32 0.002 1.00

(Fisher’s Bufonidae 6 13 0.38 0.82

Exact Test, Hylidae 18 42 0.33 0.79

P=0.0048) Leptodactylidae 15 43 0.79 0.34

Dendrobatidae 2 8 0.88 N/A

Ranidae 1 4 0.86 N/A

Centrolenidae 0 12 1.00 0.0037

Microhylidae 0 3 N/A N/A

46 Caeciliidae 0 2 N/A N/A

Rhinophrynidae 0 1 N/A N/A

Cuba Leptodactylidae 42 50 0.0728 0.9867

(Fisher’s Bufonidae 5 7 N/A 0.4351

Exact Test, Hylidae 0 1 N/A N/A

p=0.03175) Ranidae 0 1 N/A N/A

Ecuador Centrolenidae 18 26 0.0035 0.999

(Fisher’s Bufonidae 25 40 0.0054 0.998

Exact Test, Leptodactylidae 86 162 0.0003 1.000 p=2.56x10 -10 ) Plethodontidae 3 7 0.633 0.664

Dendrobatidae 12 33 0.825 0.287

Ranidae 1 4 N/A N/A

Hylidae 20 97 1.000 0.000

Caeciliidae 0 9 1.000 0.0064

Microhylidae 0 9 1.000 0.0056

Rhinatrematidae 0 2 N/A N/A

Pipidae 0 1 N/A N/A

Japan Salamandridae 2 3 N/A N/A

(Fisher’s Hynobiidae 8 17 0.163 0.948

Exact Test, Ranidae 10 22 0.141 0.948

p=0.05388) Rhacophoridae 0 8 1.000 N/A

Bufonidae 0 4 N/A N/A

Hylidae 0 2 N/A N/A

Cryptobranchidae 0 1 N/A N/A

Microhylidae 0 1 N/A N/A

47 Madagascar Microhylidae 20 41 0.0053 0.999

(Fisher’s Mantellidae 35 123 0.904 0.166

Exact Test, Hyperoliidae 0 11 1.000 0.0147

p=0.00354) Ranidae 0 2 N/A N/A

Malaysia Bufonidae 13 33 0.038 0.986

(Fisher’s Megophryidae 10 26 0.079 0.968

Exact Test, Rhacophoridae 15 42 0.061 0.975 p=0.0001903) Microhylidae 4 25 0.927 0.178

Ranidae 3 51 1.000 0.0001

Queensland, Rheobatrachidae 2 2 N/A N/A Aus

(Fisher’s Microhylidae 6 15 0.220 0.9143

Exact Test, Myobatrachidae 7 24 0.544 0.651

P=0.3207) Hylidae 11 45 0.834 0.297

Limnodynastidae 4 18 0.821 0.376

Bufonidae 0 1 N/A N/A

Ranidae 0 1 N/A N/A

Spain Discoglossidae 2 8 0.190 N/A

(Fisher’s Salamandridae 1 10 0.722 N/A

Exact Test, Ranidae 1 11 0.762 N/A

P=0.839) Bufonidae 0 4 1.000 N/A

Hylidae 0 2 1.000 N/A

Pelodytidae 0 2 1.000 N/A

Pelobatidae 0 1 N/A N/A

Sri Lanka Rhacophoridae 47 56 0.000 1.000

48 (Fisher’s Ichthyophiidae 2 3 N/A N/A

Exact Test, Bufonidae 5 8 N/A 0.484 p=7.49x10 -6) Microhylidae 4 10 N/A 0.048

Ranidae 5 15 1.000 0.0031

Uruguay Bufonidae 3 9 0.0245 N/A

(Fisher’s Hylidae 1 15 0.832 N/A

Exact Test, Leptodactylidae 0 17 1.000 N/A p=0.09145) Caeciliidae 0 1 N/A N/A

Microhylidae 0 1 N/A N/A

49 Table 2.3: Results of site level Fisher’s exact test analysis, and subsequent analysis of distribution of threatened species between amphibian families after the omission of DD species.

>Expected < Expected Species Species Family threat-level threat-level threatened Number p-value p-value

Kinabalu Megophryidae 6 13 0.0328 N/A

(Fisher’s Bufonidae 4 10 0.1505 N/A

Exact Test, Rhacophoridae 6 22 0.3575 0.8263

p=0.006557) Ranidae 1 23 0.9992 0.0092

Microhylidae 0 8 1 N/A

Mount

Nlonako Astylosternidae 8 13 0.0008 N/A

(Fisher’s Ranidae 3 10 0.3632 N/A

Exact Test, Bufonidae 2 8 0.777 N/A

p=0.01129) Arthroleptidae 2 11 0.5747 N/A

Petropedetidae 2 13 0.819 N/A

Hyperoliidae 1 24 0.9991 0.0123

Pipidae 0 3 1.000 N/A

Caeciliidae 0 2 N/A N/A

Rhacophoridae 0 1 N/A N/A

La Selva Plethodontidae 2 3 0.0358 N/A

(Fisher’s Leptodactylidae 2 13 0.4985 N/A

Exact Test, Hylidae 1 13 0.8625 N/A

p=0.4121) Centrolenidae 0 4 1 N/A

50 Bufonidae 0 3 1 N/A

Dendrobatidae 0 2 1 N/A

Ranidae 0 2 1 N/A

Caecilidae 0 1 N/A N/A

Microhylidae 0 1 N/A N/A

Australian Myobatrachidae 2 2 N/A N/A

Wet

Tropics

(Fisher’s Hylidae 5 12 0.5838 0.7059

Exact Test, Microhylidae 5 12 0.5877 0.7034

p=0.2918) Limnodynastidae 0 3 N/A N/A

Ranidae 0 1 N/A N/A

51 Table 2.4: Combined p-values obtained from Fisher’s method, illustrating the tendency for families to be over- or under-threatened within different countries.

Family χ 2 (combined p- d.f. p Direction of

value) deviation from

From Fisher’s random

Method

Leptodactylidae 21.9 8 0.0050 over-threatened

Rhacophoridae 24.0 6 0.0005 over-threatened

Bufonidae 32.3 12 0.0012 over-threatened

Ranidae 35.4 8 2.2703x10 -5 under-threatened

Hylidae 16.9 6 0.0096 under-threatened

Microhylidae 20.1 10 0.0285 under-threatened

52 Appendix

Appendix 2.1: Site species check-lists that were not suitable for analysis.

Andreone, F., 1994, The amphibians of Ranomafana rain forest, Madagascar

– preliminary community analysis and conservation considerations, Oryx ,

28(3):207-215.

Andreone, F., Randrianirina, J.E., Jenkins, P.D., and Aprea, G., 2000,

Species diversity of Amphibia, Reptilia, and Lipotyphla (Mammalia) at

Ambolokopatrika, a rainforest between the Anjjanaharive-Sud and Marojejy massifs, NE Madagascar, Biodiversity and Conservation , 9: 1587-1622.

Andreone, F. et al. , 2003, The amphibians and reptiles of Nosy Be (NW

Madagascar) and nearby islands: a case study of diversity and conservation of an insular fauna, Journal of Natural History , 37, 17, 2119-2149.

Amiet, J.-L, 1978, Les amphibians anoures de la region de Mamfé

(Cameroun), Annales Faculte des Sciences de Yaoundé , 20:33-107.

AmphibiaWeb Ecuador, Cuyabeno Reserve, Ecuador, http://www.puce.edu.ec/zoologia/vertebrados/amphibiawebec/index.html

AmphibiaWeb Ecuador , Yasuni National Park, Ecuador, http://www.puce.edu.ec/zoologia/vertebrados/amphibiawebec/index.html

53 Behangana, M., 2004, The distribution and diversity of amphibian fauna of

Lake Nabugabo and surrounding areas, African Journal of Ecology , 42, 6-13.

Behangana, M., 2004, The diversity and status of amphibians and reptiles in the Kyoga Lake Basin, African Journal of Ecology , 42, 51-56.

Braun, P.C., & Braun, C.A.S., 1980, Lista previa dos anfibios do Estado de

Rio Grande so Sul, Brasil, Iheringia , (56):121-146

Burger, M., Branch, W.R., & Channing, A., 2004, Amphibians and reptiles of

Monts Doudou, Gabon: Species turnover along an elevational gradient,

California Academy of Sciences Memoir , 28:145-186

Conservation International RAP, Noel Kempff Mercado National Park, http://www.conservation.org/xp/CABS/research/rap/results/results.xml

Conservation International RAP, Sanctuario Nacional Pampas del Heath, http://www.conservation.org/xp/CABS/research/rap/results/results.xml

Conservation International RAP, Sirenia Field Station, Costa Rica, http://www.conservation.org/xp/CABS/research/rap/results/results.xml

Duellman, W.E., Lynch, J.D.,1988, Anuran Amphibians of the Cordillera de

Cutucu, Ecuador, Proc. Acad. Nat. Sci. Philadelphia , 140(2), 125-142.

54 Frétey, T., & Dewynter, M., 1998, Amphibiens anoures de la forêt des Abeilles

(Gabon), Journal of African Zoology , 112:171-184.

Fugler, C.M. & Walls, A.B., 1979, The Anura (Amphibia) of Rio Upano Valley of eastern Ecuador, Journal of the Tennessee Academy of Science , Vol

54(4):149-155.

Heyer, W.R., Rand, A.S., Cruz, C.A.G., Peixoto, O.L. and Nelson, C.E. (1990)

Frogs of Boracéia. Arquivos de Zoologia 31: 231-410.

Hoogmoed, M.S. and Ávila-Pires, T.C.S. (1991) Annotated checklist of the herpetofauna of Petit Saut, Sinnamary River, French Guiana. Zoologische

Mededelingen Leiden 65: 53-88.

Hoogmoed, M.S., & Gorzula, S.J. 1979 Checklist of the savanna inhabiting frogs of the El Manteco region with notes on their ecology and the description of a new species of treefrog (Hylidae, Anura), Zoologische Mededelingen ,

54(13):183-211

Ibàñez, R., Rand, A.S., Jaramillo, C.A.,

1999, The amphibians of Barro Colorado Nature Monument, Soberania NP and adjacent areas.

Jatun Sacha, Ecuador, check-list from the Fundacion Jatun Sacha

55 Kiew B. H., 1987, An annotated checklist of the herperofauna of Ulu Endau,

Johore, Malaysia, Malayian Nature Journal , 41:413-423

Kok, P.J.R., 2000, A survey of the anuran fauna of Montagne Belvédère, county of Saül, French Guiana: field visit with comments on taxonomy and ecology, British Herpetological Society Bulletin , 71:6-29.

Krishnamurthy, S.V., 2003, Amphibian Assemblages in undisturbed and disturbed areas of Kudremukh National Park, central Western Ghats, India,

Environmental Conservation , 30(3), 274-282.

Largen, M.J., & Dowsett-Lemaire, F., 1991, Amphibians (Anuran) from the

Kouilou River Basin, République de Congo, Tauraco Research Report , 4:145-

168.

LeBreton, M., 1999, A working checklist of the herpetofauna of Cameroon, with localities for species occurring in southern Cameroon and a list of herpetofauna for the Dja Faunal Reserve, Netherlands Committee for IUCN.

Lips K.R., Reeves, J.D., Witters, L.R., 2003, Ecological Traits Predicting

Amphibian Population Declines in Central America, Conservation Biology ,

17(4) :1078-1088. Amphibian species list for El Cope National Park, Costa

Rica.

56 McCracken, S., Cocha Cashu Biological Station, Manu National Park, Peru, check-list from of www.tadpoleorg.org

McCracken, S., Department of Loreto, Peru, from of www.tadpoleorg.org

McCracken, S., Rio Ituxi, Brazil, from www.tadpoleorg.org

McCracken, S., Tiputini Biodiversity station, Ecuador, www.tadpoleorg.org

Mijares-Urrutia, A., & Arends, A.R., 2000, Herpetofauna of Estado Falcon, northwestern Venezuela: A checklist with geographical and ecological data.

Smithsonian Herpetological Information Service, No.123.

Nussbaum, R.A., Raxworthy, C.J., Raselimanana, A.P., Ramanamanjato,

J.B., Amphibians and Reptiles of the Reserve Naturelle Integrale d’Andohahela, Madagascar. Fieldiana Zoology , New Series, No. 94,

(1999):155-173.

Ohler, A., Marquis, O., Swan, S. and Grosjean, 2000, S.Amphibian biodiversity of Hoang Lien Nature Reserve (Lao Cai province, northern

Vietnam) with description of two new species, Herpetozoa , 13(1-2) 71-87.

Oklahoma Museum of Natural History, Roraima, Brazil, http://www.omnh.ou.edu/personnel/herpetology/NSF/DEB_9200779/

57 Oklahoma Museum of Natural History, Sucumbios, Para Province, Brazil, from http://www.omnh.ou.edu/personnel/herpetology/NSF/DEB_9200779/

Pérez Peña, P. & Curno,O., Reserva Nacional Allapahuayo-Mishana, Peru, check-list from, of Wildlife Conservation Society, Peru

Pérez Peña, P., & Curno, O., Reserva Nacional Pacaya Samiria, Peru, check-list from of Wildlife Conservation Society, Peru.

Raselimanana, A.P., Raxworthy, C.J., & Nussbaum, R.A., Herpetofaunal

Species Diversity and Elevational Distribution Within the Parc

National de Maraojejy, Madagascar, Fieldiana Zoology , New Series, No. 97

(2000):157-174.

Raxworthy, C.J., & Nussbaum, R.A., (1996), A Floral and Faunal Inventory of the Eastern Slopes of the Reserve Naturelle Integrale d’Andringitra,

Madagascar: With Reference to Elevational Variation. Fieldiana Zoology ,

New Series, No. 85:158-170.

Raxworthy, C.J., Andreone, F., Nussbaum, R.A., Rabibisoa, N.,

Randriamahazo, H., 1998, Amphibians and Reptiles of the Anjanaharibe-Sud

Massif, Madagascar: Elevational Distribution and Regional Endemicity.

Fieldiana Zoology, New Series, No. 90:79-92.

58 Rödel, M.-O. & R. Ernst, 2003, The Amphibians of Marahoué and Mont Péko

National Parks, Ivory Coast, Herpetozoa , 16(1-2):23-39.

Rödel, M.O., & Agyei A.C., Amphibians of the Togo-Volta highlands, eastern

Ghana, Salamandra , 39(3/4):207-234

Rödel, M.O., 2003, The amphibians of Mont Sangbé National Park, Ivory

Coast, Salamandra, 39(2): 91-110.

Rödel, M.O., & Branch, W.R., 2003, Herpetological survey of the Haute Dodo and Cavally forests, western Ivory Coast, Part 1: Amphibians, Salamandra ,

38(4):245-268.

Rödel, M.O., Bangoura, M.A., & Böhme, W., 2004, The amphibians of south- eastern Republic of Guinea, Herpetozoa , 17(3/4): 99-118.

Rödel, M.O., Gil, M., Agyei, A.C., Leache, A.D., Diaz, R.E., Fujita, M.K., &

Ernst, R., 2005, The amphibians of the forested parts of south-western

Ghana. Salamandra , 41(3):107-127.

Suarez-Mayorga, A.M., 1999, Lista preliminary de la fauna amphibian presente en el transecto la Montañita-Alto de Gainbete, Coquitá, Colombia,

Revista de la Academia Colombiana de Ciencas Exactas Fisicas y Naturales,

23:395-405

59 Chapter 3: Predicting susceptibility to future declines in the world’s frogs

The manuscript presented in this chapter is published as:

Bielby, J., Cooper, N., Cunningham, A.A., Garner, T.W.J., & Purvis, A., (In

Press), Conservation Letters ,

Abstract

The 2004 Global Amphibian Assessment demonstrated that almost 400 anuran species have recently moved closer to extinction due to a host of threat mechanisms. Of particular concern is the role of the fungal pathogen,

Batrachochytrium dendrobatidis (Bd ), for which more traditional conservation management is not effective. Determining which biological and environmental factors affect a species’ susceptibility to these mechanisms would greatly aid conservation prioritisation and planning. I performed phylogenetic comparative analyses to determine which biological and environmental factors predict species’ susceptibility to rapid declines, both generally and in the context of Bd . My results extend the findings of previous finer scale studies: we find that high-altitude, restricted-range, aquatic species with low fecundity are most likely to suffer Bd -related declines. I use my findings to identify those species most at risk of Bd -related declines and global extinction in the future, and identify areas where many species are predicted to be susceptible.

Identifying susceptible species in advance of their decline is particularly important in setting priorities when, as here, declines are hard to arrest once underway.

60 Introduction

In recent decades, amphibian species have experienced serious population declines on every continent on which they are found (Blaustein and Wake

1990; Houlahan et al. 2000; Wake 1991; Wyman 1990). The geographic extent and severity of amphibian population declines make the subject a conservation issue of high priority. Amphibian extinction risk and population declines are taxonomically non-random (Bielby et al. 2006; Stuart et al.

2004), even at small geographic scales (Bielby et al. 2006; Lips et al. 2003;

Williams and Hero 1998), indicating that species biological attributes influence susceptibility to decline. Understanding patterns in intrinsic susceptibility is therefore an important step towards identifying locations with many species susceptible to decline in future (Cardillo et al. 2006), and in tailoring conservation management to the needs of particular species.

Previous analyses on susceptibility to population declines in amphibians suggest that large size, restricted range, living at high altitude, ecological specialism, and low-fecundity may all contribute to susceptibility (Lips et al.

2003; Murray and Hose 2005; Williams and Hero 1998), but these studies have generally considered local declines rather than species-level dynamics

(but see La Marca et al. 2005; Sodhi et al. 2008). If global conservation management decisions are to be directed by research (Mendelson et al.

2006b) we need to ensure that the information supporting decisions on the allocation of conservation resource is robust, and that species most at risk of future decline are identified. Knowing which type of species are the most

61 likely to become globally more threatened would therefore be a powerful tool to guide the allocation of conservation resource.

The GAA (IUCN et al. 2004), provided the first evaluation of all known amphibian species according to the Red List criteria. Stuart et al. (2004) described a total of 398 anuran species that had experienced a genuine increase in extinction risk. These species, termed “Rapid Decline” (hereafter

RD) species, may represent some of the most urgent conservation problems

(Stuart et al. 2004) – species whose total population has declined to the point that they have moved a step closer to global extinction. The status changes of

RD species were attributed to three causes: habitat reduction, over- exploitation, and enigmatic decline. While the former two categories require little explanation, the third is more complex. “Enigmatic decline” species have declined for reasons that are not fully understood. Possible mechanisms include climate change, UV-B irradiation, pollution, and infectious disease

(Collins and Storfer 2003; Stuart et al. 2004). In particular, many of the species categorised as enigmatic RDs are thought to have declined due to chytridiomycosis (e.g. Atelopus species, La Marca et al. 2005), caused by the fungal pathogen, Batrachochytrium dendrobatidis (hereafter Bd ).

Current understanding (Daszak et al. 1999; Lips et al. 2003; Murray and Hose

2005; Sodhi et al. 2008; Williams and Hero 1998) makes several predictions about how biology and threat processes interact to dictate anuran susceptibility to rapid declines. Using a large, geographically widespread dataset, and controlling for phylogeny, I aimed to test some of these

62 predictions, particularly in the context of Bd related declines. I analysed RDs to find rules-of-thumb about which biological and environmental traits make species more susceptible to increased extinction risk in the future, and to form a basis for recommendations on future conservation efforts (Cardillo et al.

2006).

Specifically, we addressed the following hypotheses:

a) Restricted range species will be more susceptible to rapid decline as

any threatening process is more likely to affect the entire distribution.

Species with small ranges will also tend to have small population sizes

and so may be more susceptible to stochasticity (Brown 1995).

b) Large species will be more susceptible to rapid decline. In many taxa,

large body size correlates with traits that elevate extinction risk or

likelihood of population decline such as low population density (Damuth

1987) or slow life history (McKinney 1997; Pimm et al. 1988). Large

anurans, particularly those with small ranges, have previously been

shown to have increased probability of local population decline (Lips et

al. 2003). Global extinction risk is also higher in large species in other

taxa (Bennett and Owens 1997; Cardillo et al. 2005).

c) Species with low reproductive rate and/or fecundity will be more

susceptible to decline as they will be less able to compensate for

increased mortality (MacArthur and Wilson 1967; McKinney 1997;

Purvis et al. 2000b).

d) High-altitude anurans will be more susceptible, as these species often

have restricted ranges and may be exposed to environmental

conditions that exacerbate certain threat processes such as

63 chytridiomycosis (Bosch et al. 2007; Daszak et al. 1999; Laurance et

al. 1996). Additionally, high altitude species tend to have slower life

histories (Morrison and Hero 2003; Morrison et al. 2004). e) Species that rely on freshwater habitats for one or more stages of their

lifecycle are likely to experience a wider range of threats (e.g. pollution,

infectious disease) than strictly terrestrial species, and are therefore

more likely to decline. Additionally as Bd is an aquatic organism,

species reliant on freshwater habitats are more likely to be exposed to

the pathogen. f) According to the chytrid-thermal-optimum hypothesis (Pounds et al.

2006), species living in environmental conditions favourable to

propagation of Bd (Piotrowski et al. 2004) will be more likely to have

experienced Bd -related declines (Berger et al. 2004; Bosch et al.

2007). Additionally, species limited to a narrow range of environmental

conditions may have a lower tolerance to change and thus could be

more susceptible to RDs due to environmental peturbation (McKinney

1997; Purvis et al. 2000b). g) Species exposed to higher human populations densities may be more

likely to have suffered RD as they are more likely to experience habitat

degradation and loss (but see Balmford 1996; Cardillo et al. 2004), and

are perhaps more likely to be overexploited.

64

Methods

Data

As 91% (398/435) of amphibian RD species are anurans (frogs), and the biology of the three amphibian orders are so different, I concentrated the analyses solely on frogs. The final dataset contained 553 anuran species representing 32 of the 33 anuran families, and containing species from all six continents on which amphibians are present. Data were collected from a range of sources, including peer-reviewed literature, grey literature, field guides, web sites, and direct contact with GAA Red List species assessors (a full list of sources used is included in appendix 3.1). When data were collected for a species in a given geographic location, I collected data from species of a range of statuses (i.e. non threatened, threatened, and RD) from that location in order to reduce the probability that our results would merely reflect geographic heterogeneity in threat intensity. I biased our data collection towards RD species in order to maximise power to test our hypotheses. The percentage of RD species in the final dataset (41.0%,

227/553) was therefore much higher than the percentage of RD species in all anuran species (7.9%, 398/5066). Of the 227 RD species in the dataset, 15

(6.6%) were over-exploited, 77 (33.9%) were threatened by habitat reduction, and 134 (59.0%) were enigmatic decline species ( c.f. 11%, 42.1%, and 47.6% respectively of RD species as a whole). One species was threatened by both over-exploitation and enigmatic decline. Of the 326 non-RD species in the dataset, 194 (59.5%) were non-threatened, and 132 (40.5%) were threatened

65 (c.f. 57.8% non-threatened, 42.2% threatened respectively in anurans as a whole after exclusion of DD species).

In order to determine which biological, environmental and anthropogenic factors predicted species’ RD status, I collected information on the following twelve variables using the taxonomy of Frost (2004) throughout:

Geographic ranges (in km 2) and aquatic life-stage data came from the GAA

(IUCN et al. 2004) database. A species was classified as aquatic life-stage if it relied on freshwater for any stage (e.g. reproduction, egg deposition, development) of its life cycle.

Mean snout-vent length (mm) was used as a measure of species body size, and mean clutch size (eggs per clutch) was taken as a measure of a species’ speed of life history. Where I had more than one value of a variable for a species, I used the mean. In order to investigate the role of a species’ environment on RD status, I obtained representative values for a species using geographic distribution data (IUCN et al. 2004) and spatial datasets of altitude (m) (Hijmans et al. 2005), annual actual evapotranspiration (AET)

(mm) (Willmott 2001), net primary productivity (NPP) (mm) (Willmott 2001), isothermality (a measure of annual temperature consistency), maximum temperature of the warmest month , precipitation seasonality , and precipitation in the driest quarter (Hijmans et al. 2005). To examine whether human impacts played a significant role in RDs, I used a map of human population density (people/km 2) (CIESIN. 2000). To extract the spatial data required from the GIS data layers, I overlaid shape files of species geographic ranges

66 on a 30-second grid and calculated the median grid cell value within the geographic range.

I transformed explanatory data to meet requirements of normality where necessary: geographic range (km 2), snout-vent length (mm), eggs per clutch, altitude, and human population density were all log transformed, and precipitation of the driest quarter was square-root transformed; other variables remained untransformed.

Analyses

I coded species according to their IUCN status, RD status and causal mechanism (using information from the GAA), and whether Bd infection has been reported ( i.e. Bd +/Bd -) (see appendix 3.2 for list of diagnosed species of which I am aware, and of references used). This information allowed us to make four binary comparisons, addressing four questions as follows

A) Do RD species differ biologically from other threatened species?

B) Do Enigmatic RD species differ biologically from other threatened and RD species?

C) Do Bd + RD species differ biologically from Bd + species that have not suffered a rapid decline?

D) Do RD species that have been infected/diagnosed as Bd + differ biologically from RD species not infected/diagnosed with Bd ?

When comparing threatened and non-threatened species, there are two possible problems of circularity. First, threat status might be autocorrelated

67 with geographic range size, because species may be classified as threatened based on their narrow distribution. This could also bias comparisons between

RD species (which are by definition at least Near-Threatened) and non-RD species (many of which are Least Concern). Most of my tests are structured to compare RD species with other RD or threatened species, minimising this circularity. The remaining analysis (comparison C above) also included Least

Concern species in order to maximise the sample of Bd infected species. To remove the circularity from this analysis, I omitted threatened species (RD or otherwise) that had not been categorised under IUCN Criterion A (range decline, rather than small range size), following Purvis et al. (2000). The second possible problem is that RD status could be further autocorrelated with geographic range size if estimates of decline rate are artefactually higher in small-range species. This could occur if such species are easier to monitor, or if small ranges are mapped at a finer scale than large ranges (Thomas and

Abery 1995). However, this artefact would predict that RD species tend to have smaller ranges than other threatened species, which they do not (see results of comparison A, below).

Due to the binary nature of the response variables, and the consequent non- normal error structure, I conducted the analyses using generalized linear models specifying the link function as either logit or complementary log-log, and preferred the link with the lowest residual deviance (Crawley 2002).

We accounted for the strong phylogenetic signal within the dataset (see appendix 3.3) using generalised estimating equations (GEE) as described by

68 Paradis and Claude (2002). GEE are a procedure in which model parameters in a generalised linear model framework may be estimated taking correlations among observations into account. All GEE analyses were conducted using the compar.gee function (package = ape; Paradis et al. 2005) in the software package R (R Development Core Team 2005). In my analyses, the correlation matrix describing the relationships among data was obtained from the recently published genus-level amphibian phylogeny (Frost et al. 2006) with taxonomy as a surrogate for phylogeny below the genus level. As a result of the lack of branch length information, all branch lengths were set to two units. For each of the four comparisons, A-D, I initially conducted single-predictor analyses for each of the twelve explanatory variables. In two of the 48 bivariate analyses performed, the model parameters would not converge when using one or both of the possible link functions. I have indicated where this occurred in the results section. To account for the possibility that the large number of predictors tested would result in an increased rate of Type I errors, the Holm-

Bonferroni method was implemented on the bivariate analyses within each of the four comparisons made. All Holm-Bonferroni adjustments were made using the p.adjust function in the software package R (R Development Core

Team 2005).

Once bivariate relationships had been modelled, I investigated multipredictor relationships, allowing determination of which variables were associated with

RD status when correlated variables were accounted for. Initially, all significant single predictors were included in the model, and the predictor with the highest p-value was successively removed until the remainder were

69 significant. To check the robustness of the resulting model, each previously excluded explanatory variable was added in turn until no more significant predictors could be found. Additionally, I conducted the analyses including variables that did not converge as single predictors in the first step of the model building process. If no significant predictors were found in the bivariate analyses, I built a minimum adequate model by including all predictors in the first model and dropping those with the highest p-values until all remaining terms had a p-value below 0.05.

For the multipredictor models obtained, cross-validation was used to obtain a measure of model fit. For each of 1000 iterations, six species from the original model were randomly selected, and brier scores calculated using the observed and fitted values of those species. Brier scores (mean squared error for binary data (Brier 1950)) are a measure of model fit: a model with perfect predictive ability would obtain a score of 0.0, while a score of 1.0 would indicate a model with very poor predictive power. For all multipredictor models, I calculated a mean brier score and standard error of the mean (SEM) for the 1000 iterations.

Using the parameters from my most predictive model, that of Bd -related RD

(comparison C), and the relevant biological trait data I calculated the probability of decline for 3976 anuran species. To illustrate those locations with a high number of highly susceptible species, I mapped all species with a probability of RD of 1.0 if infected with Bd (figure 3.1) using ArcMap 9.0.

70 Results

When RD species were compared with other threatened species (comparison

A, Table 3.1), rapid declines were associated with aquatic life-stage in bivariate comparisons. The multipredictor model included aquatic life-stage as a significant predictor, with large geographic range approaching a significant relationship with RD (p=0.061) (brier score 0.29 ±0.002).

Compared with threatened species, RD species affected by enigmatic decline

(comparison B) were found to live in locations with low annual temperature variation, and to have aquatic life-stages in bivariate analyses. In this multipredictor model low annual temperature variation was no longer a significant term, whereas clutch size was now significant, and high altitude approached significance with a p-value of 0.09 (Table 3.2, brier score 0.26

±0.003). When only Bd + species were analysed (comparison C), RD was associated with high altitude and small geographic range in the bivariate analyses (Table 3.2). The multipredictor analysis also included aquatic lifestage Bd + species alongside the significant bivariate predictors, as being more likely to have suffered an RD (Table 3.3, brier score 0.06 ±0.002).

When only RD species were included (comparison D) in bivariate analyses, no predictor variables were significantly associated with a species’ Bd status, though large range RD species in areas of low actual evapotranspiration were more likely to be Bd + in the multipredictor model (Table 3.4).

71 Discussion

These global analyses of RD anuran species strengthen and extend the scope of the findings of finer scale studies of anuran declines (Lips et al.

2003; Murray and Hose 2005; Williams and Hero 1998). Species with aquatic life-stages are particularly susceptible to rapid declines, while species with restricted distribution at high elevations are most susceptible to Bd related

RD. I discuss these results and their possible implications for conservation.

Narrow geographic distribution is an important predictor of high extinction risk and declines in many taxonomic groups (Cardillo et al. 2005a; Fisher and

Owens 2004; Purvis et al. 2000b) including amphibians (Cooper et al. 2008;

Murray and Hose 2005; Sodhi et al. 2008). However, the role of geographic distribution in predicting RD status appears to be more complex. While restricted range species were indeed more susceptible to decline in one of my analyses (comparison C), two of the other comparisons (A and D) showed the opposite. This observed change of role of geographic distribution may be explained by the epidemiology and particularly the spread of chytridiomycosis.

Whereas wider ranging species are more likely to be infected and diagnosed with Bd , it is restricted range species that are more likely to suffer serious consequences of chytridiomycosis.

One of the three criteria for disease-induced extinction is a restricted distribution, and associated small population size (de Castro and Bolker

2005). While wide-ranging species become infected (comparison D), and even suffer local declines (e.g. Alytes obstetricans (Bosch et al. 2001) ,

72 Salamandra salamandra (Bosch and Martinez-Solano 2006)), they are more likely to recover or at least persist through recruitment, recolonisation, and immigration than restricted range species. Of the 133 confirmed and possible amphibian extinctions in recent years (IUCN et al. 2004; Schloegel et al.

2006), many of which are likely to have involved chytridiomycosis, the majority of species had restricted ranges.

Aside from geographic range size, which other factors make a species more susceptible to the deleterious effects of Bd infection? My analyses of both enigmatic RD (comparison B) and Bd infected species (comparison C) strongly support regional work on biological traits associated with amphibian population declines in Central America and Australia (Daszak et al. 1999;

Hero et al. 2005; Lips et al. 2003; Murray and Hose 2005; Williams and Hero

1998): high altitude, small clutch size, partially aquatic species are most susceptible. Caution may be required when using these variables to predict

RD: species with traits that place obvious restrictions on a distribution (e.g. high elevation, or aquatic life-stage) are perhaps more likely to be classified as RD simply because of a higher certainty of changes to the distribution of those species. However, there are strong biological mechanisms that could link aquatic life-stage and high altitude to RD (Lips et al. 2003; Williams and

Hero 1998). Aquatic life-stage species are reliant on more than one type of habitat for completion of their life-cycle, and therefore may be exposed to a wider range of threats likely to cause increased mortality. One such threat would be Bd , whose aquatic life-cycle may mean that amphibians with an aquatic life-stage are more likely to be infected than those species that do not.

73 Given such an increased level of mortality, one way for a population to remain stable or recover is through high recruitment in subsequent reproductive events. This may explain the importance of clutch size as a predictor of enigmatic RD with species with larger clutch size being less likely to suffer an

RD.

The inclusion of high altitude as a correlate of Bd -related decline, even accounting for geographic distribution size, suggests that altitude does not increase a species’ susceptibility to RD simply by limiting the species’ distribution. As with restricted range species, high altitude species may be less able to offset mortality through recruitment, recolonisation, or colonisation from other populations. Additionally, factors such as Bd , climate change and pollution, may exert their influence most seriously at mid- to high-elevations

(Daly et al. 2007; Pounds et al. 2006), and could perhaps work in combination to affect amphibian populations. The decline of sympatric species of reptiles and birds (which are unaffected by Bd ) at sites of amphibian declines

(Pounds et al. 1999; Whitfield et al. 2007), underlines the difficulty involved in ascribing an ultimate cause to a decline (Collins and Storfer 2003; Mace and

Balmford 2000). Similarly, declines of Bd -infected amphibian populations as a result of climatic conditions rather than disease (Daszak et al. 2005) highlight the fact that the presence of Bd does not necessarily mean that it is responsible for declines in the area, something that should be remembered when interpreting the analyses presented here. So, while chytridiomycosis is a major cause for concern in amphibian conservation, the effects of other

74 enigmatic factors, either alone or in combination with infectious diseases, must not be ignored.

The suitability of predictive models for conservation prioritisation depends upon the strength of their predictive ability, which varied greatly in the models presented here. Comparison of Bd + species (comparison C) resulted in a model with extremely high predictive ability (brier score of 0.064 ±0.002) compared to the other models, particularly those for comparisons A and B

(brier scores of 0.29 ±0.002 and 0.26 ±0.003 respectively). The key to the lower predictive ability of those models may lie with the greater geographic and taxonomic coverage in the respective datasets. The data on Bd infection were more limited geographically, taxonomically and in sample size than the data forming the basis of the other comparisons, which therefore reduced the variation in the dataset. However, given a suitable level of predictive ability, such models may be of use in guiding conservation priorities.

Using those traits highlighted in the analyses we may be able to direct conservation research and policy by predicting which species are most susceptible in the future (e.g. Cardillo et al. 2006; Koh et al. 2004). I have used the model parameters obtained in my most predictive model, comparison C, to estimate the probabilities that each of 3976 anuran species would decline rapidly if they became infected with RD (for list of species fitted values, see Bielby et al. Conservation Letters, In press ). Figure 3.1 shows a map of the locations of those species with an estimated probability of RD of 1, and identifies areas which may be particularly heavily hit if Bd were to be

75 introduced into the region (figure 3.1). Given the susceptibility of their species, these locations represent a high priority for pre-emptive conservation policy and management efforts. I echo the call of Sodhi et al (2008) for multi- foci management, which may include monitoring species population trends

(Collins and Halliday 2005); directing more detailed local studies (e.g. Bosch et al. 2007; Pounds et al. 2006; Williams and Hero 1998); screening for pathogens; the introduction of legislation to reduce the possibility of pathogen introduction via the amphibian trade (Fisher and Garner 2007); preventing the spread of Bd by establishing field hygiene protocols for people working in the area (e.g. Australian Government 2006); possibly the establishment of ex-situ populations (Mendelson et al. 2006b); and the development of national abatement plans in countries in which Bd is already known to be present (e.g.

Australian Government 2006). While some of the locations highlighted are already the subject of Bd -related declines and associated conservation attention (e.g. high altitudes in Latin America; Young et al. 2001) , and others are already high profile recipients of conservation efforts (e.g. Madagascar), other locations (e.g. New Guinea and the Western Ghats) have received little, if any, amphibian conservation attention or management so far.

Of the 837 species with a predicted probability of RD of 1 in the event of Bd infection, 385 are listed as data deficient in the IUCN Red List. The data deficient status of so many susceptible species highlights the need for better basic information on a large proportion of anuran species, including improving our knowledge of population status and trends, particularly in Asia and Africa

(Collins and Halliday 2005).

76

In order to maximise the utility and accuracy of such models we could incorporate more detailed information on both predictor and response variables into the models. Better knowledge of anuran life-history and ecology, and more information on Bd distribution and pathogenicity of different strains

(Berger et al. 2005b) would add greatly to the prediction of response to Bd infection. Likewise, using long-term species population data, rather than solely using coarse measures such as change of Red List status, would add further resolution to predictive modelling of species response to threatening processes, and would represent a useful future avenue for research on the subject of amphibian population declines (Collins and Halliday 2005).

Although my results are generally consistent with the results of finer-scale studies, there were some differences. My analyses, for example, did not identify any climatic predictors, or any effect of species’ habitat preference on probability of decline, in contrast to finer-scale studies (Burrowes et al. 2004;

Daszak et al. 2005; Pounds et al. 2007; Williams and Hero 1998). The advantages of studying single species or regions are the ability to include such detailed information, which is unlikely to be available globally, and the ability to inform focussed conservation goals (Fisher and Owens 2004).

However, detailed studies on single species suggest heterogeneity among populations in likelihood of infection and development of disease (e.g. Alytes obstetricans ; S. Walker pers. comm), and results of regional studies may also vary within a clade (e.g. Hero et al. 2005; e.g. Lips et al. 2003; Williams and

Hero 1998). These differences may reflect local idiosyncrasies in the exact causal mechanism of decline. So while single species and regional studies

77 have the ability to incorporate more detailed information, their findings may not be applicable to wider sets of species and locations: large-scale studies are also an important tool in identifying species’ susceptibility to threat mechanisms.

As the IUCN systematically updates Red List assessments for a range of taxa, the amount of information on species status over time will increase, and so will our ability to look at trends and patterns in species’ status. Determining susceptibility of species to declines in the future could complement schemes monitoring changes in Red list status (Butchart et al. 2007; Butchart et al.

2006), help to set priorities for future conservation actions, and offer a useful alternative to concentrating conservation efforts solely upon species already experiencing a high risk of extinction (e.g. Ricketts et al. 2005).

78 Tables

Table 3.1: Results of univariate and multipredictor analyses of comparison A (RD species vs threatened species). N/A indicates that the analysis of this variable could not be completed for computational reasons related to the function compare.gee. dfp = phylogenetic degrees of freedom. † log geographic range could only be analysed using the clog-log function. In the MAM reported below, n = 357, link = clog-log, deviance = 102.74, dfp = 49.95, brier score = 0.29 S.E.+/- 0.0022.

bivariate analyses MAM Bonferroni n link t deviance t p-value adj. p-value log geographic range 359 clog-log† 2.05 0.43 111.53 1.92 0 .06 log snouth vent length 358 logit 1.09 1.00 81.22 log eggs per clutch N/A N/A N/A N/A N/A log altitude 359 logit -0.99 1.00 83.69 log human population density 358 logit 0.33 1.00 83.70 Net primary productivity 331 clog-log -0.48 1.00 78.34 Actual evapotranspiration 354 clog-log 1.05 1.00 83.54 Isothermality 359 clog-log -0.48 1.00 84.90 Temp of warmest month 359 clog-log 0.44 1.00 85.73 Precipitation seasonality 359 clog-log 0.07 1.00 84.33 √ precipitation of driest quarter 359 clog-log -0.15 1. 00 84.20 Aquatic lifestage 357 logit 3.90 0.00 82.02 3.06 0.00

79 Table 3.2: Results of univariate and multipredictor analysis of comparison B (enigmatic RD species vs threatened species). In the MAM reported below, n = 120, link = clog-log, deviance = 30.02, dfp = 21.53, brier score = 0.26 S.E.+/- 0.003

bivariate analyses MAM Bonferroni n link t deviance t p-value adj. p-value log geographic range 359 clog-log -0.80 1.00 87.81 log snouth vent length 358 clog-log -0.42 1.00 87.66 log eggs per clutch 106 logit -2.70 0.10 28.23 -2.569 0.02 log altitude 359 clog-log 0.78 1.00 88.36 1.77 0.09 log human population density 358 logit -0.43 1.00 87.05 net primary productivity 331 logit -0.51 1.00 82.35 actual evapotranspiration 354 clog-log 0.10 1.00 88.69 isothermality 359 clog-log 3.18 0.02 80.41 temp of warmest month 359 logit -0.71 1.00 86.35 precipitation seasonality 359 logit 1.50 1.00 91.10 √ precipitation of driest quarter 359 logit -0.41 1.00 88.73 0.01 0.04 aquatic lifestage 357 clog-log 2.49 87.35 2.121

80 Table 3.3: Results of univariate and multipredictor analysis of comparison C ( Bd + RD species vs all other Bd + species). † log eggs per clutch could only be analysed using the clog-log function. In the MAM reported below, n = 102, link = clog-log, deviance = 7.37, dfp = 16.94, brier score = 0.06 S.E.+/-0.002

bivariate analyses MAM Bonferroni n link t adj. p-value deviance t p-value log geographic range 102 logit -4.37 0.01 13.44 -2.65 0.02 log snout vent length 101 cloglog -0.85 1.00 21.14 log eggs per clutch 83 cloglog † -1.80 0.81 16.97 log altitude 102 logit 3.71 0.02 15.16 2.65 0.02 log human population density 102 cloglog 1.19 1.00 20.10 net primary productivity 102 logit -1.18 1.00 19.25 actual evapotranspiration 102 cloglog -0.84 1.00 19.53 isothermality 102 cloglog 1.17 1.00 25.79 temp of warmest month 102 logit -2.04? 0.58 16.41 precipitation seasonality 102 cloglog 1.16 1.00 20.57 √ precipitation of driest quarter 102 cloglog -0.83 1.00 20.45 aquatic lifestage 102 logit 0.69 1.00 19.66 2.12 0.05

81 Table 3.4: Results of univariate and multipredictor analysis of comparison D ( Bd + RD species vs Bd - RD species). N/A indicates that the analysis of this variable could not be completed for computational reasons related to the function compare.gee. In the MAM reported below, n, = 227, link = logit, deviance = 32.83, dfp = 33.82, brier score = 0.14 S.E.+/- 0.003

bivariate analyses MAM Bonferroni n link t adjusted p-value deviance t p-value log geographic range 227 logit 2.48 0.19 34.05 3.24 0.003 log snout vent length 226 logit 2.64 0.14 34.55 log eggs per clutch 79 logit 1.88 0.72 18.83 log altitude 227 clog-log 0.48 1.00 47.05 log human population density 227 N/A N/A 1.00 N/A net primary productivity 227 clog-log -0.61 1.00 40.84 actual evapotranspiration 227 logit -0.34 1.00 42.64 -3.001 0.0052 isothermality 227 clog-log -0.90 1.00 38.91 temp of warmest month 227 clog-log -0.45 1.00 44.75 precipitation seasonality 227 clog-log -0.06 1.00 44.10 √ precipitation of driest quarter 227 clog-log -1.62 0.92 35.71 aquatic lifestage 227 clog-log 0.66 1.00 43.96

82 Figure Figure 3.1: Global distribution of anuran species with a predicted probability of Bd -related decline =1. Geographic ranges of those species with a predicted probability of 1 are marked in red.

83 Appendices

Appendix 3.1: References used in anuran data collection.

AmphibiaWeb : Information on amphibian biology and conservation, 2005. Berkeley, California: AmphibiaWeb. http://amphibiaweb.org/ . (Accessed May- August 2005).

Anstis, M. 2002. Tadpoles of South Eastern Australia. A guide with keys. Sydney: New Holland Publishers (Australia) Pty Ltd.

Ashton Jr., R.E., and Ashton, P.S. 1988. Handbook of reptiles and amphibians of Florida. Part three: the amphibians. Miami: Windward Publishing Inc.

Barbour, and Noble, 1920, Bull. Mus. Comp. Zool., 63: 413

Barker, J., Grigg, G.C., and Tyler, M.J. 1995. A field guide to Australian frogs. Chipping Norton (NSW): Surrey Beatty and Sons Pty Ltd.

Bell, B.D. 1978. Observations on the ecology and reproduction of the New Zealand leiopelmid frogs. Herpetologica, 34: 340-354.

Bennett, R. 1997. Reptiles and frogs of the Australian Capital Territory. Woden: National Parks Association of the ACT Inc.

Biju, S.D. and Bossuyt, F. 2003. New frog family from India reveals an ancient biogeographical link with the Seychelles. Nature, 425, 711–714.

Blommers-Schlösser, R.M.A, and Blanc, C.P. 1991. Faune de Madagascar 75: Amphibiens. Paris: Museum National d’Histoire Naturelle.

Campbell, J.A. 1998. Amphibians and reptiles of Northern Guatemala, the Yucatán, and Belize. Norman: University of Oklahoma Press.

84

Cannatella, D. 1995. Allophryne ruthveni. The Tree of Life Web Project, http://tolweb.org/tree?group=Allophryne_ruthveni&contgroup=Neobatrachia . (Accessed May- August 2005).

Cannatella, D.C., Ford, L., and Bockstanz, L. 1997. Salientia: frogs and toads. The Tree of Life Web Project. www.tolweb.org/tree?group=Salientia . (Accessed May- August 2005).

Channing, A., Amphibians of Central and Southern Africa, Cornell University Press.

Cei, J.M. 1980. Amphibians of Argentina. Monograph 2. Monitore Zoology Italia.

Chanda, S.K. 2002. Handbook-Indian amphibians. Calcutta: Director of ZSI.

Cochran, D.M. 1954. Frogs of south-eastern Brazil. Washington D.C.: Smithsonian Institution Press.

Collins, J.T., Amphibians and Reptiles in Kansas, Kansas: Museum of Natural History, Kansas.

Coloma, L.A., and Lötters, S. 1996. The tadpole of Atelopus balios (Anura: Bufonidae) from the Pacific lowlands of Ecuador. Herpetologica, 52: 66-70.

Coloma, L.A., Lötters, S., and Salas, A.W. 2000. Taxonomy of the Atelopus ignescens complex (Anura: Bufonidae): designation of a neotype Atelopus ignescens and recognition of Atelopus exiguus. Herpetologica, 56: 303-324.

Conant, R., and Collins, J.T. 1991. Peterson field guides: Reptiles and amphibians of eastern/central North America. 3 rd edn. Boston: Houghton Mifflin Company.

85 Davies, M., and Richards, S.J. 1990. Developmental biology of the Australian hylid frog Nyctimystes dayi (Günther). Transactions of the Royal Society of South Australia, 114: 207-211.

Del Pino, E., Personal communication regarding clutch size in Gastrotheca riobambae and G. pseutes.

Degenhardt, W.G., Painter, C.W., and Price, A.H. 1996. Amphibians and reptiles of New Mexico. Albuquerque: University of New Mexico Press.

Deuchar, E. 1975. Xenopus: The South African clawed frog. London: John Wiley and Sons Ltd.

De Witte, G.F. 1948. Faune de Belgique: amphibiens et reptiles. 2 nd edn. Bruxelles: Musée Royal d’Histoire Naturelle de Belgique.

Duellman, W.E. 1970. The hylid frogs of Middle America. Volume I and II. Monograph of the Museum of Natural History Kansas.

Duellman, W.E. Altig, R., 1978, New species of tree frogs (family Hylidae) from the Andes of Colombia and Ecuador. , Herpetologica, 34:2, 177-185.

Duellman, W.E. Hillis, D.M. 1987. Marsupial frogs (Anura: Hylidae: Gastrotheca) of the Ecuadorian Andes: resolution of taxonomic problems and phylogenetic relationships, Herpetologica, 43:2, 141-173.

Duellman, W.E. 2005. Cusco Amazónico. The lives of amphibians and reptiles in an Amazonian rainforest. New York: Cornell University Press.

Dundee, H.A., Rossman, D.A., The Amphibians and Reptiles of Louisiana, Baton Rouge, Louisiana State University Press.

Dutta, S.K., and Manamendra-Arachchi, K. 1996. The Amphibian Fauna of Sri Lanka. Columbo: Wildlife Heritage Trust of Sri Lanka.

86

Flores, G. 1985, A new Eleutherodactylus from the Amazonian slopes of the Ecuadorian Andes, Herpetologica, 41(4):447-450

Fouquette, M.J. 1969. Rhynophrynus dorsalis. Catalogue of American reptiles and amphibians. Washington D.C.: American Society of Icthyologists and Herpetologists.

Franklin,C.J. 2002. The Surinam Toad. The Cold Blooded News: The Newsletter of the Colorado Herpetological Society, 29: Volume 11. http://coloherp.org/cb-news/Vol-29/cbn-0211/Surinam.php . (Accessed May 2005).

Glaw, F., and Vences, M. 1994. A fieldguide to the amphibians and reptiles of Madagascar. 2 nd edn. Leverkusen: Moos Druck..

Hedges, S.B., Personal communication regarding clutch size in Eleutherodactylus patricae and E.pituinus.

Henkel, F.W., and Schmidt, W. 2000. Amphibians and reptiles of Madagascar and the Mascarene, Seychelles and Comoro islands. Malabar, Florida: Krieger Publishing Company.

Heselhaus, R., Poison arrow frogs: their natural history and care in captivity. Blandford Press, London.

Heyer, W.R., and Crombie, R.I. 1979. Natural history notes on Craspedoglossa stejnegeri and Thoropa petropolitana (Amphibia: Salientia; Leptodactylidae). Journal of the Washington Academy of Sciences, 69: 17-20.

Inger, R.F. 1954. Systematics and Zoogeography of Phillipine amphibians. Chicago Natural History Museum.

87 Inger, R.F. 1966. The systematics and zoogeography of the amphibians of Borneo. Chicago: Field Museum Press.

IUCN/ISSG Global Invasive Species Database. 2005. http://www.issg.org/database . (Accessed May 2005).

Jacobson, S.K., and Vandenberg, J.J. 1991. Reproductive ecology of the endangered Golden toad (Bufo periglenes). Journal of Herpetology, 25: 321- 327.

Kenny, J.S. 1969. The Amphibia of Trinidad. Reprinted from Studies on the fauna of Curaçao and other Caribbean islands No.108. (ed. P.W. Hummelinck). The Hague: Martinus Nijhoff.

Kuzmin, S.L. 1999. The amphibians of the former Soviet Union. Sofia: Pensoft Publishers.

Lambiris, A.J.L. 1989. The frogs of Zimbabwe. Torino: Museo Regionale di Scienze Naturali.

Le Berre, F. 1993. Notes on three species of frogs of the genus Mantella. The Vivarium, 4: 19-22.

Lee, J.C. 1996. The amphibians and reptiles of the Yucatán Peninsula. Ithaca: Comstock Publishing Associates.

Liu, C. 1950. Amphibians of Western China. Chicago: Field Museum Press.

Lötters, S. 1996. The neotropical toad genus Atelopus. Checklist-Biology and Distribution. Köln: Verlags GbR.

Lutz, B. 1973. Brazilian species of Hyla. Austin: University of Texas Press.

88 Lynch, J.D. 1986. Notes on the reproductive biology of Atelopus subornatus. Journal of Herpetology, 20: 126-129.

Matz, G., and Weber, D. 1983. Guide des amphibiens et reptiles d’Europe. Paris: Delachaux et Niestlé.

Merino-Viteri, A. Personal communication regarding clutch size in Telmatobius niger.

Meyer, J.R., and Foster, C.F. 1996. A Guide to the frogs and toads of Belize. Malabar, Florida: Kreiger Publishing Company.

Minter, L.R., Burger, M., Harrison, J.A., Braack, H.H., Bishop, P.J., and Kloepfer, D. (eds) 2004. Atlas and red data book of the frogs of South Africa, Lesotho and Swaziland. SI/MAB Series #9. Washington D.C.: Smithsonian Institution Press.

Nöllert, A., and Nöllert, C. 2001. Amfibieëngids van Europa. Baarn: Tirion Uitgevers.

Norman, D.R. 1994. Amphibians and reptiles of the Paraguayan Chaco Volume I. San José: RARE Center.

Palis, J.G. 1998. Breeding biology of the Gopher frog, Rana capito, in Western Florida. Journal of Herpetology, 32: 217-223.

Rabor, D.S. 1981. Phillipine reptiles and amphibians. Quezon City: Abiva Publishing House.

Robb, J. 1986. New Zealand Amphibians and reptiles. Auckland: Collins.

Rueda-Almonacid, J.V., Lynch, J.D., and Amézquita, A. (eds) 2004. Libro rojo de los anfibios de Colombia. Bogata: Conservacion Internaciónal Columbia.

89

Salvador, A. 1986. Guia de campo de los anfibios y reptiles de la peninsula Iberica, Isles Baleares y Canarias. Madrid: Santiago Garcia.

Salvador, A., and Paris, M.G. 2001. Anfibios Españoles: Identificación, historia natural y distribución. Talavera de la Reina: Canseco Editores S.L.

Samollow, P.B. 1980. Selective mortality and reproduction in a natural population of Bufo boreas. Evolution, 34: 18-39.

Savage, J.M. 2002. Amphibians and reptiles of Costa Rica: a herpetofauna between two continents, between two seas. University of Chicago Press.

Schleich, H.H., and Kästle, W. (eds). 2002. Amphibians and reptiles of Nepal. Koenigstein: Koeltz Scientific Books.

Schwartz A., 1964, Caribbean Journal of Science, 4:473-484

Schwartz, A. 1965. A new species of Eleutherodactylus (Amphibia:Leptodactylidae) from the Sierra de Baorucor Republics Dominican. Proc. Biol. Soc.Wash. 78:165–168

Schwartz, A., and Henderson, R.W. 1991. Amphibians and reptiles of the West Indies. Descriptions, distributions and natural history. Gainsville: University of Florida Press.

Stebbins, R.C. 1985. A field guide to western reptiles and amphibians. 2 nd edn. Boston: Houghton Mifflin Company.

Terent’ev, P.V., and Chernov, S.A. 1965. Key to amphibians and reptiles. 3 rd edn. Jerusalem: Israel Program for scientific translations.

90 Trueb, L. 1979. Leptodactylid frogs of the genus Telmatobius in Ecuador, with the description of a new species. Copeia 1974:714-733

Vredenberg, V. 2000. Rana muscosa: Egg predation. Herpetological Review, 31: 170-171.

Wake, M.H. 1978. The reproductive biology of Eleutherodactylus jasperi (Amphibia, Anura, Leptodactylidae), with comments on the evolution of live bearing systems. Journal of Herpetology, 12: 121-133

Wright, A.H., and Wright, A.A. 1949. Handbook of frogs and toads of the United States and Canada. Ithaca: Comstock Publishing Company Inc.

91 Appendix 3.2: List of Bd+ species Family Binomial Reference

Hylidae Acris crepitans Pessier, A.P.,et al. (1999). Cutaneous chytridiomycosis in poison dart frogs ( Dendrobates spp.) and White's tree frogs ( Litoria caerulea ). Journal of Veterinary Diagnostic Investigation 11:194-199.

Myobatrachidae Adelotus brevis URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Ranidae Afrana fuscigula Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221. Discoglossidae Alytes obstetricans Bosch, J., I. Martinez-Solano, and M. Garcia-Paris. 2001. Evidence of a chytrid fungus infection involved in the decline of the common midwife toad ( Alytes obstetricans ) in protected areas in central Spain. Biological Conservation 97:331-337 Ambystomatidae Ambystoma Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen macrodactylum Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221. Ambystomatidae Ambystoma tigrinum Davidson D, et al. Chytridiomycosis in Arizona (USA) tiger salamanders. Getting the Jump! on amphibian disease: Conference and workshop compendium. Cairns, 26-30 August 2000;23. Bufonidae Atelopus bomolochos La Marca, E.,et al2005. Catastrophic population declines and extinctions in neotropical harlequin frogs (Bufonidae: Atelopus ). Biotropica 37:190-

92 201

Bufonidae Atelopus carbonerensis La Marca, E.,et al2005. Catastrophic population declines and extinctions in neotropical harlequin frogs (Bufonidae: Atelopus ). Biotropica 37:190- 202 Bufonidae Atelopus chiriquiensis La Marca, E.,et al2005. Catastrophic population declines and extinctions in neotropical harlequin frogs (Bufonidae: Atelopus ). Biotropica 37:190- 203 Bufonidae Atelopus cruciger La Marca, E.,et al2005. Catastrophic population declines and extinctions in neotropical harlequin frogs (Bufonidae: Atelopus ). Biotropica 37:190- 204 Bufonidae Atelopus mucubajiensis La Marca, E.,et al2005. Catastrophic population declines and extinctions in neotropical harlequin frogs (Bufonidae: Atelopus ). Biotropica 37:190- 205 Bufonidae Atelopus pulcher La Marca, E.,et al2005. Catastrophic population declines and extinctions in neotropical harlequin frogs (Bufonidae: Atelopus ). Biotropica 37:190- 206 Bufonidae Atelopus sorianoi La Marca, E.,et al2005. Catastrophic population declines and extinctions in neotropical harlequin frogs (Bufonidae: Atelopus). Biotropica 37:190- 207

93 Bufonidae Atelopus spumarius La Marca, E.,et al2005. Catastrophic population declines and extinctions in neotropical harlequin frogs (Bufonidae: Atelopus ). Biotropica 37:190- 208 Bufonidae Atelopus varius La Marca, E.,et al2005. Catastrophic population declines and extinctions in neotropical harlequin frogs (Bufonidae: Atelopus ). Biotropica 37:190- 209 Bufonidae Atelopus zeteki Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Plethodontidae Bolitoglossa colonnea Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Plethodontidae Bolitoglossa schizodactyla Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Bombinatoridae Bombina pachypus Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221.

94 Bufonidae Bufo boreas Muths et al., Biological Conservation 110 (2003) 357–365

Bufonidae Bufo bufo Bosch and Solano, 2006, Oryx Vol 40 No 1 January 2006 Bufonidae Bufo canorus Carey, C., Cohen, N. and Rollins-Smith, L. A. 1999. Amphibian declines: An immunological perspective. Developmental Comparative Immunology 23: 459-472. Bufonidae Bufo coniferus Lips, K. R., F. Brem, R. Brenes, J. D. Reeve, R. A. Alford, J. Volyes, C. Carey et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165-3170. Bufonidae Bufo haematiticus Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221.

Bufonidae Bufo marinus URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Centrolenidae Centrolene ilex Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Centrolenidae Centrolene prosoblepon Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165-

95 3170.

Centrolenidae Cochranella albomaculata Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Centrolenidae Cochranella euknemos Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Dendrobatidae Colostethus flotator Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Dendrobatidae Colostethus nubicola Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Dendrobatidae Colostethus panamensis Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the

96 National Academy of Sciences of the United States of America 103:3165- 3170. Dendrobatidae Colostethus pratti Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Dendrobatidae Colostethus talamancae Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor azueroensis Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor bransfordii Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor bufoniformis Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the

97 National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor cerasinus Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor crassidigitus Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor fitzingeri Puschendorf, R., Bolanos, F., Chaves, G., 2006, The amphibian chytrid fungus along an altitudinal transect before the first reported declines in Costa Rica, Biological Conservation 132:136-142 Leptodactylidae Craugastor gollmeri Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor megacephalus Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165-

98 3170.

Leptodactylidae Craugastor noblei Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor podiciferus Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor punctariolus Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor tabasarae Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Craugastor talamancae Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the

99 National Academy of Sciences of the United States of America 103:3165- 3170.

Myobatrachidae Crinia georgiana URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Crinia glauerti URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Crinia insignifera URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Crinia pseudinsignifera URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Dendrobatidae Dendrobates auratus Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Dendrobatidae Dendrobates pumilio Puschendorf, R., Bolanos, F., Chaves, G., 2006, The amphibian chytrid fungus along an altitudinal transect before the first reported declines in Costa Rica, Biological Conservation 132:136-142 Dendrobatidae Dendrobates vicentei Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Hylidae Duellmanohyla uranochroa Puschendorf, R., Bolanos, F., Chaves, G., 2006, The amphibian chytrid fungus along an altitudinal transect before the first reported declines in

100 Costa Rica, Biological Conservation 132:136-142

Hylidae Ecnomiohyla miliaria Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Eleutherodactylus Lips, K. R. et al. 2006. Emerging infectious disease and loss of caryophyllceus biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Eleutherodactylus cruentus Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Eleutherodactylus Lips, K. R. et al. 2006. Emerging infectious disease and loss of diastema biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Eleutherodactylus emcelae Berger, L., R. Speare, P. Daszak, D. E. Green, A. A. Cunningham, C. L. Goggin, R. Slocombe et al. 1998. Chytridiomycosis causes amphibian

101 mortality associated with population declines in the rain forests of Australia and Central America. Proceedings of the National Academy of Sciences of the United States of America 95:9031-9036. Leptodactylidae Eleutherodactylus Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen melanostictus Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221. Leptodactylidae Eleutherodactylus Lips, K. R. et al. 2006. Emerging infectious disease and loss of museosus biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Eleutherodactylus patriciae Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221. Leptodactylidae Eleutherodactylus pituinus Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221. Leptodactylidae Eleutherodactylus ridens Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Eleutherodactylus vocator Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165-

102 3170.

Leptodactylidae Gastrotheca cornuta Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Gastrotheca pseustes Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221.

Myobatrachidae Geocrinia rosea URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Geocrinia vitellina URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Helioporus australiacus URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Helioporus barycragus URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Helioporus eyrei URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Centrolenidae Hyalinobatrachium Lips, K. R. et al. 2006. Emerging infectious disease and loss of colymbiphyllum biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Hylidae Hyla arenicolor Sredl, Michael and Dennis Caldwell. (2000). Wintertime Population Surveys - Call for Volunteers. Sonoran Herpetologist (Tucson

103 Herpetological Newsletter) 13:1.

Hylidae Hyla regilla Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221. Hylidae Hyla vasta Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221. Hylidae Hylomantis lemur Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Hylidae Hyloscirtus colymba Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Hylidae Hyloscirtus palmeri Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Hylidae Istmohyla pseudopuma Puschendorf, R., Bolanos, F., Chaves, G., 2006, The amphibian chytrid fungus along an altitudinal transect before the first reported declines in

104 Costa Rica, Biological Conservation 132:136-142

Myobatrachidae Lechriodus fletcheri URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Leiopelmatidae Leiopelma archeyi Kingsley D. Environment News, 23 April 2002. http://www.abc.net.au/science/news/enviro/EnviroRepublish_537533.htm., Bell 2004 Leptodactylidae Leptodactylus Lips, K. R., F. Brem, R. Brenes, J. D. Reeve, R. A. Alford, J. Volyes, C. pentadactylus Carey et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165-3170.

Myobatrachidae Limnodynastes dorsalis URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Limnodynastes dumerilii URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Myobatrachidae Limnodynastes tasmaniensis URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Myobatrachidae Limnodynastest terraereginae URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria adelaidensis URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria aurea URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria barringtonensis URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

105 Hylidae Litoria booroolongensis URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria caerulea URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria chloris URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria citropa URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria ewingii URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria genimaculata URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria gracilenta URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria infrafrenata URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria lesueuri URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria moorei URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria nannotis URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria nasuta URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria pearsoniana URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria peronii URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria phyllochroa URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria raniformis URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

106 Hylidae Litoria rheocola URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria spenceri URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Litoria verreauxii URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Mixophyes balbus URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Mixophyes fasciolatus URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Mixophyes fleayi URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Mixophyes iteratus URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Bufonidae Nectophrynoides asperginis Weldon and du Preez, Froglog, number 62, April 2004, Microhylidae Nelsonophryne aterrima Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170.

Myobatrachidae Neobatrachus pelobatoides URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Hylidae Nyctimystes dayi URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Plethodontidae Oedipina collaris Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165-

107 3170.

Plethodontidae Oedipina parvipes Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Dendrobatidae Phyllobates lugubris Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170. Leptodactylidae Physalaemus pustulosus Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170.

Myobatrachidae Pseudophryne corroboree URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Pseudophryne pengilleyi URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Salamandridae Pseudotriton ruber Don Nichols (2002) pers com Ranidae Ptychadena anchietae Speare R, Berger L. Global distribution of chytridiomycosis in amphibians. World Wide Web - http://www.jcu.edu.au/school/phtm/PHTM/frogs/chyglob.htm. 11

108 November 2000.

Hylidae Ptychohyla hypomykter Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221. Ranidae Rana arvalis Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221. Ranidae Rana berlandieri Sredl, Michael and Dennis Caldwell. (2000). Wintertime Population Surveys - Call for Volunteers. Sonoran Herpetologist (Tucson Herpetological Newsletter) 13:1. Ranidae Rana blairi Sredl, Michael and Dennis Caldwell. (2000). Wintertime Population Surveys - Call for Volunteers. Sonoran Herpetologist (Tucson Herpetological Newsletter) 13:1. Ranidae Rana catesbeiana Mitchell JC, Green DE. Chytridiomycosis in two species of ranid frogs in the southeastern United States. Joint Meeting of the American Society of Ichthyologists and Herpetologists, Herpetologists' League, and Society for the Study of Amphibians and Reptiles, 4-8 July 2002, Kasas City, USA.

Ranidae Rana chiricahuensis Morell, V. (1999). Are pathogens felling frogs? Science 284: 728-731. Ranidae Rana pipiens Carey, C., Cohen, N. and Rollins-Smith, L. A. 1999. Amphibian declines: An immunological perspective. Developmental Comparative Immunology 23: 459-472.

109 Ranidae Rana sphenocephala Mitchell JC, Green DE. Chytridiomycosis in two species of ranid frogs in the southeastern United States. Joint Meeting of the American Society of Ichthyologists and Herpetologists, Herpetologists' League, and Society for the Study of Amphibians and Reptiles, 4-8 July 2002, Kasas City, USA. Ranidae Rana vibicaria Puschendorf, R., Bolanos, F., Chaves, G., 2006, The amphibian chytrid fungus along an altitudinal transect before the first reported declines in Costa Rica, Biological Conservation 132:136-142 Ranidae Rana warszewitschii Lips, K. R. et al. 2006. Emerging infectious disease and loss of biodiversity in a Neotropical amphibian community. Proceedings of the National Academy of Sciences of the United States of America 103:3165- 3170.

Ranidae Rana yavapiensis Morell, V. (1999). Are pathogens felling frogs? Science 284: 728-731.

Salamandridae Salamandra salamandra Bosch and Martinez-Solano, 2006, Oryx Vol 40 No 1 January 2006 Ranidae Strongylopus grayii Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221.

Myobatrachidae Taudactylus acutirostris URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm

Myobatrachidae Taudactylus eungellensis URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Leptodactylidae Telmatobius niger Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-221.

110 Myobatrachidae Uperoleia laevigata URL: http://www.jcu.edu.au/school/phtm/PHTM/frogs/chy-au-status.htm Pipidae Xenopus gilli Weldon, C., L. H. du Preez, A. D. Hyatt, R. Muller, and R. Speare. 2004. Origin of the amphibian chytrid fungus. Emerging Infectious Diseases 10:2100-2105 Pipidae Xenopus laevis Weldon, C., L. H. du Preez, A. D. Hyatt, R. Muller, and R. Speare. 2004. Origin of the amphibian chytrid fungus. Emerging Infectious Diseases 10:2100-2106 Pipidae Xenopus muelleri Weldon, C., L. H. du Preez, A. D. Hyatt, R. Muller, and R. Speare. 2004. Origin of the amphibian chytrid fungus. Emerging Infectious Diseases 10:2100-2107

111 Appendix 3.3: Results of ANOVAs used to determine whether there were significant differences between anuran families in the explanatory variables used in the analyses. Significant results indicate the presence of a phylogenetic structure in the data. *p<0.05, **p<0.01, ***p<0.001

Variable d.f. F-statistic Log geographic range 30 and 522 7.28***

Log snout-vent length 30 and 521 9.10***

Log eggs per clutch 25 and 224 9.73***

Log altitude 30 and 522 3.07***

Log human population 30 and 521 7.13*** density

Net primary productivity 29 and 495 3.08***

Net actual evapotrans 28 and 519 9.94***

Isothermality 30 and 522 12.35***

Mean temperature of the 30 and 522 3.72*** warmest month

Precipitation seasonality 30 and 522 5.09***

Square root of the 30 and 522 4.18*** precipit ofdriest quarter

Freshwater dependence 30 and 520 22.98***

112 Chapter 4: Modelling extinction risk in multispecies data sets: phylogenetically independent contrasts vs. decision trees

Abstract

In recent years there has been an increase in the number of studies of extinction risk attempting to determine what differences exist between threatened and non-threatened species. One potential problem in such studies is the existence of phylogenetic non-independence that species-level data may contain. However, the use of phylogenetic comparative methods

(PCM) to account for non-independence remains controversial, and some recent studies of extinction have recommended the use of other methods that do not account for phylogenetic non-independence, notably decision trees

(DTs). Here I perform a systematic comparison of techniques, comparing the performance of PCM regression models with corresponding non-phylogenetic regressions and DTs over different clades and response variables. I found that predictions made were broadly consistent among techniques, but that predictive precision varied across techniques with PCM regression and DTs performing best. Despite their inability to account for phylogenetic non- independence, DTs were useful in highlighting interaction terms for inclusion in the PCM regression models. We discuss the implications of these findings for future studies of extinction risk.

113 Introduction

Given the current loss of biodiversity that we face, there is a real need for ecology and conservation to become more predictive in order to make predictions that practitioners or policy-makers can use (e.g. Sutherland 2006).

One area in which predictive models have become more common in recent years is in the examination of species’ susceptibility to the processes causing increased levels of extinction risk. In particular, there has been an increased interest in how inherent traits, such as species life-history and ecology, and external factors, such as human population pressure, interact to produce the current pattern of species risk. As a result the number of studies looking for correlates of extinction risk has grown considerably, incorporating a wide range of taxonomic groups (Cardillo et al. 2005a; Cooper et al. 2008; Koh et al. 2004; Laurance 1991; Owens and Bennett 2000; Reed and Shine 2002;

Sullivan et al. 2000) at a range of scales (Fisher and Owens 2004; Purvis et al. 2005a) . These studies have two main aims. First, they can discriminate among competing hypotheses about which intrinsic traits of species’ biology predispose species to decline in the face of anthropogenic impacts. Second, quantitative models linking biological traits to extinction risk can be used to predict which species may face a high risk of extinction in the future, and therefore inform conservation management options (Cardillo et al. 2006;

Cardillo et al. 2004; Jones et al. 2006; Sullivan et al. 2006) .

With these aims in mind, many comparisons between threatened and non- threatened species have been made. However, although extinction risk itself

114 is not an evolved trait (Cardillo et al. 2005b; Putland 2005) , it often has a phylogenetic signal (Bennett and Owens 1997; Bielby et al. 2006; McKinney

1997; Purvis et al. 2000a; Russell et al. 1998) as do many of the proposed predictor variables (Freckleton et al. 2002; Purvis et al. 2005a). Ignoring this phylogenetic signal and treating species level information as independent data is likely to result in pseudoreplication (Felsenstein 1985) which may render subsequent analyses statistically invalid, with an elevated Type I error rate (Harvey and Pagel 1991; although Type II errors may also occur

Gittleman and Luh 1992). In order to avoid such errors, the use of phylogenetically independent contrasts (Felsenstein 1985), or other similar methods such as generalised estimating equations (Paradis and Claude

2002), has become widespread when testing proposed correlates of extinction risk (Fisher and Owens 2004; Purvis 2008) .

Although their use is widespread, there are questions about the suitability of phylogenetic comparative methods (PCM) in studies of extinction risk.

Although extinction risk often shows a phylogenetic signal, the signal is not necessarily as strong as assumed by the most commonly used approach, the independent contrasts method (Collen et al. 2006; see also Cardillo et al.

2005a; Purvis et al. 2005a) . The correction for non-independence may therefore be too severe, potentially leading to ‘over-correction’ (Ricklefs and

Starck 1996) , with younger nodes having undue influence on subsequent analyses (Purvis 2008). While some studies have incorporated data-led branch length adjustments in order to at least ameliorate over-correction (e.g.

Cardillo et al. 2005a; Halsey et al. 2006; Stuart-Fox et al. 2007) these may

115 not remove the problem entirely. Further error may be introduced into PCM analyses as a result of inaccuracies in the phylogenetic tree, branch lengths, and violations of assumptions made regarding the mode of evolution (Ives et al. 2007; Ricklefs and Starck 1996; Symonds 2002) limiting the benefits of using PCM. Additionally, the sheer number of candidate predictor variables included in models of extinction risk may potentially remove the need to account for phylogeny: if a model contains most of the important variables that produce non-independence in the error term of the model, the need to correct for phylogenetic relationships can be greatly reduced (Grafen 1989).

Although categorical predictor and response variables are frequently included in analyses (Fisher and Owens 2004) , most PCM approaches are designed to deal with continuous traits. Some methods (e.g. Grafen 1989) permit predictors to be continuous or discrete, but very few (e.g. Paradis and Claude

2002) are intended for binary response variables (e.g., threatened vs. not threatened), which may be important in studies of extinction risk, but remain problematic (Purvis 2008; Read and Nee 1995).

The early focus in PCM was testing whether an association between two variables was significant when pseudoreplication was removed (Felsenstein

1985; Ridley 1983), more than investigating the form of the relationship (e.g. nonlinearity) or estimating parameters. Extinction risk models are often complex, with many hypothesised predictors potentially interacting, and parameter estimation is now more commonly of interest as PCM are being used to build predictive models with the aim of, for example, prioritising

116 conservation actions (Cardillo et al. 2006). Although phylogenetic and nonphylogenetic approaches estimate the same underlying parameters (Pagel

1993) , measurement error of the traits used has a greater effect on phylogenetic than nonphylogenetic comparative studies (Harmon and Losos

2005; Purvis and Webster 1999; Ricklefs and Starck 1996). Additionally, most implementations of PCM assume linearity of relationships and do not lend themselves to straightforward testing of this assumption. Although complexities such as non-linearities and interactions between explanatory variables can be included in the environment of a model incorporating PCM

(e.g. Cardillo et al. 2005a; Cooper et al. 2008) , more commonly they are not

(Quader et al. 2004). Visualising such complexities when using contrasts is more difficult than with raw species data, which perhaps explains why they are not more often included in analyses. The addition of non-linearities and interaction terms alongside numerous predictor variables may also lead to

‘parameter proliferation’ and further reduce the power of the analyses in question (Crawley 2002).

Decision trees (hereafter DTs) provide an alternative approach for building complex predictive models that may be more capable of dealing with some of these obstacles (Breiman et al. 1984; Crawley 2002; De'ath and Fabricius

2000) . DTs are statistical models that highlight which explanatory variables

(continuous or categorical) are the most important in explaining variation in a specified response variable (continuous or categorical). DTs are built by repeatedly splitting the data into two subsets. At each split, or branch, the data is partitioned on the basis of whether it falls above or below a threshold

117 value of a selected predictor variable. The predictor variable and threshold value are selected that explain most deviance in the response variable, making the two subsets as homogenous as possible. The process is continued until further splits explain no further deviance or the terminal node reaches a minimum size (i.e., number of cases). In explaining variation in the response, each variable may be used once, more than once, or not at all; this allows identification of variables that interact, have nonlinear effects, or are not included in the model. In recent years the utility of DTs has been reflected in the number of publications using them for a range of purposes, such as modelling climatic niches (e.g. Araujo et al. 2005a; Araujo et al. 2005b) and species distributions (Elith et al. 2006) . It has recently been argued that DTs are an ideal way to model extinction risk (Sullivan et al. 2006) and conservation status (Jones et al. 2006; Koh et al. 2004) but, while they are able to highlight important interactions, they do not account for the phylogenetic structure in comparative data.

If PCMs and DTs are both considered as useful tools in modelling extinction risk, it is important to clarify how the two approaches perform relative to each other and non-PCM (hereafter TIPS) regression models when analysing trait associations with extinction risk. Although Sullivan et al. (2006) advocated the use of DTs, they did not perform suitable comparable analyses using independent contrasts or an alternative PCM. Here, we analyse extinction risk (which we treated as a coarsely measured continuous response variable; see Purvis 2005a) in both amphibians and mammals, and Rapid Decline (a binary response variable) in amphibians, using PCM, TIPS regression models

118 and DTs. The use of different clades, kinds of response variable, and phylogenetic hypotheses of differing certainty allow us to compare how the modelling techniques perform under a range of conditions with the aim of evaluating the strengths of each in building predictive models of extinction risk. Specifically we aimed to address the following four questions:

1. Do DTs identify interactions and non-linearities of explanatory variables?

2. Are the results of DTs and TIPS regressions affected by the possible non- independence of species level data?

3. Are the predictions made regarding species susceptibility to extinction risk consistent across modelling techniques?

4. How precise are the predictions made by the different techniques?

By highlighting strengths and weaknesses of all three approaches and evaluating how similar their results are, we hope to be able to make recommendations for future analyses of extinction risk.

Methods

Data

Response variables

We took PCM analyses of extinction risk from three recent completed studies and applied TIPS and DT to them. Following the original analyses (Bielby et al. 2008; Cardillo et al. 2008; Cooper et al. 2008), we modelled extinction risk in two clades (mammals and frogs) and four separate orders within mammals, using both PCM and TIPS regression models, and DTs. In both mammalian

119 and anuran analyses our measure of extinction risk were the IUCN Red List categories (IUCN 2004) converted to a continuous index ranging from 0-5, 0 corresponding to “Least Concern”, 5 to “Extinct in the Wild” or “Extinct”

(Cardillo et al. 2005a; Cooper et al. 2008; Purvis et al. 2005a) . We excluded threatened species not listed under criterion A of the Red List from our analyses of extinction risk to avoid circularity.

Additionally, we modelled a binary variable, Rapid Decline (hereafter “RD”), in anurans. RD species are those that have experienced a genuine increase in

IUCN Red list category since 1980. In addition to comparing all RD species with non-RD threatened species, we analysed susceptibility to RD as a result of different threatening processes: “enigmatic” threats (Stuart et al. 2004), chytridiomycosis, and infection by the pathogen Batrachochytrium dendrobatidis (hereafter “ Bd”) among RD species. The range of comparisons made allowed us to evaluate the utility of DTs, phylogenetic comparative method (PCM), and TIPS regression models in a number of different clades, using phylogenetic trees of different certainty, with different types of response variables (i.e. binary and continuous). The exact details of the analyses performed are described in Table 4.1

Predictor variables

Anurans

Biological data on frogs species came from the dataset underlying two previous analyses (Bielby et al. 2008; Cooper et al. 2008). Full details of data collection, and variables examined are provided in Cooper et al. 2008, chapter

120 3 and appendix 3.1. The final anuran dataset contained 553 frog species, representing 32 of 33 families from all six continents on which amphibians are present. The number of species in the dataset was restricted largely by the lack of biological and life history data available.

Mammals

Biological data on mammalian species came from the “PanTheria” database, a dataset of ecological and life-history traits for 4030 mammalian species, from over 3300 published literature sources. Full details of data collection and quality checking are provided in K. Jones et al. (manuscript in preparation).

Mammalian geographic distributions (Grenyer et al. 2006; Sechrest et al.

2002) were used to calculate range size and for summarizing human impact and environmental conditions within each species’ range. Details of the predictor variables included in the mammalian analyses are in Cardillo et al .

(2008).

Analyses

Phylogenetic and non-phylogenetic regression models

Modelling extinction risk in anurans (comparison A)

In frogs, the relationship between extinction risk and the predictor variables outlined in chapter 3 were explored using phylogenetically independent contrasts as outlined in Cooper et al. (2008). A minimum adequate model

(MAM) was built by including all variables in the model and deleting those with the highest p-value until all remaining terms were significant.

121

Modelling RD in anurans (comparisons B-E).

Phylogenetic models of frog susceptibility to Rapid Decline were built using generalised estimating equations (GEEs) (Paradis and Claude 2002), as described in Bielby et al. (2008 in press ).

For all of the frog analyses, we used a recently published genus-level amphibian phylogeny (Frost 2007) with taxonomy as a surrogate for phylogeny below the genus level. As we did not have branch lengths for all species, we set branch lengths to equal 1 unit. All frog regression models can be found in appendix 4.1.

Modelling extinction risk in mammals and mammalian orders (comparisons F-

J).

In mammals as a whole (comparison F) and each of four clades (comparisons

G-J) extinction risk was modelled using phylogenetically independent contrasts using a dated, composite supertree phylogeny of 4510 mammalian species (Bininda-Emonds et al. 2007) , and following the protocol of Cardillo et al. (Cardillo et al. 2008). MAMs were built by starting with a full set of predictor variables, including additive effects only and followed the model building procedures described in Cardillo et al. (Cardillo et al. 2004) and

Purvis et al. (2000b). For all analyses using PCM, equivalent TIPS models were built using the same model-building algorhythms. Contrasts were calculated after transforming phylogenetic branch lengths by raising them to a power ( κ), with the value of κ optimized for each variable to minimize the

122 correlation between absolute scaled contrasts and their standard deviations

(Garland et al. 1992). All mammalian and mammalian clade regression models can be found in appendix 4.2.

Tree models

DTs were first of all grown to a maximal size, that is, until further splits result in no decrease in the residual deviance or terminal nodes reach a minimum size (n=5), and then pruned to an optimum that is determined by cross- validation techniques. However, as a result of the high occurrence and the uneven distribution of missing values of predictor variables among cases, we first needed to reduce the impact of missing values on the tree-building process and the results obtained. The default setting is for DTs is to leave out any cases with missing values, which may result in a tree built on a subset of the data that is limited in size by an unimportant predictor variable (i.e. one that does not explain variance in the response variable). Here we suggest a process maximising the sample size of cases included in the DT. A maximal tree was grown from the full set of predictor variables. Of the predictor variables not included in the maximal tree, the sample-limiting variable was removed from the dataset. This reduced the number of cases with missing values, and increased the number of cases entering the tree building process.

The full-sized tree was then rebuilt with the new, reduced data subset. Once again the unimportant predictor most limiting sample size was dropped from the tree building subset. This process was continued until no further changes to the full tree occurred.

123 The tree size (i.e. the number of terminal nodes) was then optimised using 10- fold cross-validation conducted on each tree size from the maximum to the minimum number of terminal nodes. For each of 500 iterations, the dataset was split into 10 subsets, each of which in turn was excluded from the tree- building process, while the remaining nine were used to model the data. The excluded subset was then used to validate the model by comparing the predicted fitted values with observed values of the response, yielding an estimated error (i.e. squared errors; SE) for that subset. The 500 iterations of the ten-fold validation process therefore led to 5000 values of SE for a tree of the specified size, which were used to calculate a mean squared error (MSE) and associated standard error. The optimal tree size was the smallest tree within one standard error of the tree size with the lowest MSE. Once the optimal tree size was known, a tree of that size was built using the dataset obtained in the sample size maximising process (De'ath and Fabricius 2000).

All DTs were built using the tree function (package = tree) of the statistical language R (R Development Core Team 2005), and all cross-validation was performed using scripting written in R.

Predictive performance

Consistency of model predictions

In order to check the consistency of the predictions made by different modelling techniques, we performed pairwise correlations of fitted values obtained using each technique. The correlations were limited to those species that were common to the two techniques being compared.

124 Precision of model predictions

Although strong correlations between fitted values may indicate that predictions are consistent across techniques, they do not indicate how precise these predictions were. In order to determine whether the prediction precision varied significantly among techniques the residual values obtained from the models were squared and compared using Kruskal-Wallis tests with modelling technique as a factor, the null hypothesis being that no significant differences existed. In the case of a significant difference among techniques, post-hoc non-parametric multiple comparison tests were conducting to determine where those differences lay. Squared residual values (squared error) were a suitable measure of predictive precision as they quantify the difference between the predicted and observed extinction risk of a given species. All

Kruskal-Wallis and subsequent post-hoc tests were conducted using the kruskal.test and npmc (library = npmc) functions of the statistical language R

(R Development Core Team 2005).

Phylogenetic structure in model residuals

The squared residual values obtained from each model were analysed using

ANOVA with taxonomic Family as a factor to determine whether a phylogenetic structure was present. The presence of a phylogenetic structure in the model residuals suggests that the predictions made varied in precision among subclades at or below the Family level.

125 Complexities in the data

In order to determine whether DTs were useful in outlining complexities in the data we compared interactions and non-linearities included as significant predictors in the regression models to those outlined in the DTs. This allowed a qualitative comparison of complexities in the data. Additionally, we also fitted in PCM regression models any important interactions highlighted by DT models, with the aim of seeing whether they were significant predictors of extinction risk, and whether they improved the predictive precision judged by the MSE. In order to fit interaction terms, the single predictors within the term were also included in the model.

Results

Optimal tree size

The optimal DT size obtained through 10-fold cross-validation for each comparison is given in table 4.1. The size of optimum trees varied greatly ranging from only two terminal nodes for comparison A to 13 terminal nodes for comparison I (for details of each optimal DT model, please see appendix

4.3).

Consistency of model predictions

The consistency of the predictions made by all the three techniques was strong and the fitted values were always significantly correlated (see Table 4.2 for summary, appendix 4.4 for full details). However, there was variation in the strength of the correlations among comparisons. In two of the four analyses with a binary response (comparisons C and E), there was a marked difference

126 in predictive consistency between methods. In these comparisons the consistency was lower between PCM and the other methods (see Table 4.2) than between TIPS regression and DTs, suggesting that non-phylogenetic approaches made more consistent predictions. However, in comparisons with a continuous response variable (A, F-J) there was generally a stronger correlation between PCM and TIPS regression than between DTs and the other methods (Table 4.2).

Precision of the model predictions

Kruskal-Wallis tests on the model residuals showed that there were significant differences in the predictive abilities of modelling techniques within a comparison. Results of non-parametric multiple comparison tests show where these differences lay (see Figure 4.1 and Table 4.2 for summary, appendix

4.5 for full details). In three comparisons, DTs predicted level of risk more precisely than PCM (comparisons B, C, and J), whereas PCM was more precise in two comparisons (A and G). DTs and PCM performed equally well in five comparisons (D, E, F, H, and I). In comparisons with a binary response

(B-E) DT performed particularly well having a predictive precision that was better or on a par with PCM (Table 4.2). In two of the six comparisons with a continuous response variable, PCM made predictions that were more precise than DT (comparisons A and G, but not F, H, I or J), whereas DT was more precise in one of these comparisons (J). TIPS regressions had significantly higher residual values, and therefore lower predictive precision, than PCM in all but one comparisons (comparison J).

127

Phylogenetic signal in model residual values

Analysis of residuals from PCM models suggest the presence of a strong phylogenetic effect at the Family level in eight of the ten comparisons made

(A, B, C, E, F, H, I, and J) (see Table 4.2 for summary, appendix 4.6 for full details). In one further comparison (G), model residuals had a phylogenetic signal with a p-value of between 0.05-0.1. The residuals of four TIPS regression models (comparisons A, B, C, and F) contained a significant phylogenetic signal at the Family level. Similarly, the residuals of the DTs in four of the comparisons contained a phylogenetic signal (comparisons B, C, F, and J). In TIPS regression, comparisons G, I and J had a phylogenetic signal that had a p-value of less than 0.1. Model residuals from all three techniques of comparison D lacked a phylogenetic signal. However, the small number of degrees of freedom in this and some other comparisons ( e.g. E, H, and I) means that these analyses may have very low power to detect a phylogenetic pattern.

Non-linear relationships

In PCM regression models, 11 non-linear relationships were identified as significant predictors of extinction risk in five comparisons (Comparison A: geographic range 2, geographic range 3, temperature 2 and temperature 3;

Comparison F: HPD 2 and Population density 2; Comparison G: adult body mass 2, geographic range 2, and HPD of the 5 th percentile 2; Comparison H: geographic range 2; Comparison I: HPD of the 5 th percentile 2). Fewer (six) significant non-linear relationships were found in TIPS regression models

128 (Comparison A: geographic range 2, geographic range 3, temperature 2, temperature 3, snout-vent length 2 and snout-vent 3). None of the non-linear relationships outlined in PCM or TIPS regression models were obvious from the corresponding DTs.

Interactions between predictor variables in regression models

In the PCM regression models, four significant interaction terms in three comparisons were found to be significant predictors of extinction risk

(Comparison F: geographic range:HPD, and population density:HPD;

Comparison G: geographic range:Actual Evapotranspiration; Comparison I: adult body mass:absolute latitude). No significant interaction terms were found in the TIPS regression models.

Inclusion of Interactions from DTs.

The interaction terms added to the PCM regression model on the basis of the

DTs are presented in Table 4.3. None of the interaction terms added to models of extinction risk/RD in anurans were significant predictors, although in comparison B the predictive precision of the model was improved by the addition of the term. In contrast, the interaction terms added to three of the four mammalian clade models (comparisons F, G, I, but not J) were significant predictors of extinction risk and improved the predictive precision of the models.

Discussion

The analyses of extinction risk and RD presented here were used to assess the utility of three different approaches: PCM, TIPS regression, and DTs.

129 Although there has been much discussion surrounding the use of these three techniques (Harvey and Rambaut 1998; Putland 2005; Sullivan et al. 2006) , these analyses represent the first systematic comparison of the methods, and have important implications for future studies.

Generally, both PCM and DTs performed well in terms of the precision of the predictions of a species’ extinction risk. DTs were more precise than PCM in three comparisons, while PCM were more precise in two comparisons. There were no significant differences between the predictive performance of PCM and DTs in five of the ten comparisons made. One thing that was clear from the assessment of predictive precision was the poor performance of TIPS analyses relative to the other techniques. In addition to their inability to account for phylogenetic non-independence, and higher rate of type I and type II error (Harvey and Rambaut 1998) TIPS regression methods are less precise in their predictions than alternative modelling techniques (either PCM or DT). We therefore cannot recommend the use of TIPS regression models in modelling extinction risk.

Although both PCM and DT performed well in terms of precision and consistency, some of the differences between the results suggest that DTs are not well so suited to modelling continuous, rather than binary, response variables. The limited range of predictions available via the DT approach may explain some of the technique’s relatively poor performance compared to

PCM in comparisons with a continuous response variable. In analyses of extinction risk, the observed values of the response variable ranged from 0-5,

130 as did the fitted values obtained via PCM and TIPS regression. However, with

DT the fitted values were restricted to a finite number of discrete values defined by the terminal nodes of that DT. The lower consistency in predictions made by DTs versus the regression techniques may in part be explained by the correlation of continuous and discrete fitted values in these comparisons. Likewise, the limited number of fitted values obtained via DTs compared to PCM reduced the precision with which DTs resulted in larger residual values in comparisons with continuous response variables leading to a reduced precision compared to PCM. Although in comparisons with a binary response variable DT fitted values are also limited to a finite number of discrete values, the range of possible fitted values is much lower, making this artefact less likely to dominate the differences in consistency and precision among techniques. Furthermore, the continuous response used in our analyses only ranges from 0-5; perhaps with a larger range there may be more of a discrepancy in the predictive precision of DTs and PCM regressions.

In addition to their relatively strong predictive ability, on the basis of our results

DTs may be a useful tool in identifying informative interactions in datasets

(De'ath and Fabricius 2000; Jones et al. 2006; Sullivan et al. 2006). The inclusion of interaction terms in PCMs on the basis of their appearance in DT resulted in interactions remaining as significant terms which improved predictive ability in three of five of our analyses of mammalian extinction risk

(comparisons F, G, and I). The inclusion and impact of these interactions, highlights how insightful DT models may be in informing PCMs with regards

131 complexities in the dataset. The fact that DTs did not identify all significant interactions and nonlinearities suggests that they should not be used as the sole method for finding such complexities, but that they could usefully supplement existing model building algorhythms.

One potential weakness that DT, or any other non-PCM, may have in modelling extinction risk is their inability to account for phylogenetic non- independence, a subject that has generated much discussion (Cardillo et al.

2005b; Harvey and Rambaut 1998; Putland 2005; Sullivan et al. 2006). The strong performance of PCM in the analyses presented here suggests that accounting for phylogeny does allow precise predictions of species’ conservation status. Additionally, the residuals obtained from PCM analyses suggest the presence of a phylogenetic signal among taxonomic Families within a comparison, a signal that was not obvious from the corresponding

TIPS and DT analyses. The existence of a phylogenetic signal in PCM analyses is perhaps not surprising. The strong signal suggests that the relationships between explanatory and response variables varied among subclades within a comparison. By design, PCM models reduce the influence of clade-specific relationships when the overall model is fitted, which in the presence of strong among-clade differences would result in the observed structure in the residual values (see Purvis et al. 2000b). Any variation in regression parameters among subtaxa may be important in conservation planning as they highlight the importance and utility of focussed taxonomic studies if predictive models are to be used for directing applied management or policy (Cardillo et al. 2008; Fisher and Owens 2004).

132

So, in the context of modelling extinction risk, what implications do our results have? They suggest that, although the predictions of all three methods are consistent, there are important differences between them. The strong predictive ability of PCM and their ability to reduce the influence of clade specific relationships in the data, provide support for the use of PCM techniques as the mainstay of future efforts model extinction risk. However, the strong predictive precision of DTs, in conjunction with their ability to highlight important interactions within the data, suggests that they can also play an important role in this field. We suggest that DTs may be used as a first step in identifying terms for inclusion in a suitable method of PCM in future efforts to predict species susceptibility to extinction.

133 Tables

Table 4.1: Description of the comparisons made and optimal tree size

(defined by the number of terminal nodes) for each as determined by ten-fold cross validation.

Comparison Type of Optimal tree Comparison reference variable size Extinction risk in frogs A Continuous 2 RD vs. non-RD threatened frogs B Binary 3 Enigmatic RD vs. non-Enigmatic C Binary RD threatened frogs 4 Bd + RD species vs. Bd+ non-RD D Binary species of frogs 3 Bd+ RD species vs. Bd - RD E Binary species frogs 4 Extinction risk in mammals F Continuous 10 Extinction risk in bats G Continuous 12 Extinction risk in marsupials H Continuous 5 Extinction risk in primates I Continuous 13 Extinction risk in rodents J Continuous 11

134 Table 4.2: Summary of technique performance in terms of consistency, precision, and presence of a phylogenetic signal.

Correlation coefficient of fitted values

Modelling Mean Squared Phylogenetic Comparison

Technique Tree Error signal in residuals PCM TIPS

PCM - 0.96*** 0.78*** 0.17 (A) F=5.03*** A TIPS - - 0.84*** 0.83 (C) F=2.47*** DT - - - 0.76 (B) F=1.45 PCM - 0.58*** 0.66*** 0.29 (B) F=20.43*** B TIPS - - 0.53*** 0.42 (C) F=4.26*** DT - - - 0.22 (A) F=3.39*** PCM - 0.40*** 0.20* 0.26 (B) F= 6.84*** C TIPS - - 0.77*** 0.37 (C) F=4.66*** DT - - - 0.23 (A) F=6.12*** PCM - 0.97*** 0.87*** 0.06 (A) F=1.52 D TIPS - - 0.86*** 0.29 (B) F=0.51 DT - - - 0.11 (A) F=1.396 PCM - 0.48*** 0.43*** 0.14 (A) F= 2.25*** E TIPS - - 0.71*** 0.40 (B) F=0.58 DT - - - 0.22 (A) F=1.05 PCM - 0.81*** 0.66*** 0.12 (A) F= 2.52*** F TIPS - - 0.70*** 0.42 (B ) F=1.55*** DT - - - 0.24 (A) F=1.89*** PCM - 0.72*** 0.55*** 0.20 (A) F= 1.59 ? G TIPS - - 0.35*** 0.42 (C) F= 1.56 ? DT - - - 0.21 (B) F= 1.19 PCM - 0.67*** 0.63*** 0.01 (A) F= 3.53*** H TIPS - - 0.75*** 0.22 (B ) F= 1.209 DT - - - 0.13 (A) F= 1.52 PCM - 0.90*** 0.75*** 0.17 (A) F= 3.38*** I TIPS - - 0.75*** 0.66 (B) F= 1.65 ? DT - - - 0.27 (A) F= 1.85 PCM - 0.71*** 0.39*** 0.39 (B) F= 3.47*** J TIPS - - 0.45*** 0.32 (B) F= 2.00 ? DT - - - 0.28 (A) F= 2.99***

135 Table 4.3: Interaction terms added to the PCM regression models based on the corresponding DT. In comparison A and H there were no informative interactions in the optimal tree to add to the PCM model.

Predictive accuracy Original Significant (MSE) after the Comparison Interaction term predictive as a term? addition of accuracy (MSE) interaction term

A N/A N/A N/A N/A

B Aquatic lifestage and NPP No 0.29 0.22

C HPD and isothermality No 0.23 0.262

C Isothermality and Temp No 0.23 0.247

D Geographic range and altitude No 0.06 0.07

E Clutch size and HPD No 0.14 0.21

E Clutch size and altitude No 0.14 0.22

F Neonatal body mass and litters per year Yes 0.12 0.02

G Adult body mass and geographic range Yes 0.2 0.08

H N/A N/A N/A N/A

I Adult body mass and geographic range Yes 0.17 0.07

J Geographic range and AET No 0.09 0.23

136 Figure

Figure 4.1: Significant differences in MSE among techniques for each comparison. Squares = PCM, Triangles = DT, Circles = TIPS.

Horizontal bars are placed between techniques where non-parametric multiple comparisons show significant differences to occur.

137 Appendices

Appendix 4.1: Frog regression models. Further explanation of which taxonomic group each comparison relates to can be found in table 4.3.

Comparison A

Phylogenetic MAM of comparison A: extinction risk in frogs. d.f. = 129, R 2 = 0.59, MSE = 0.17

Coefficient t p-value Geographic range^2 -0.07 -8.15 <0.001 Geographic range^3 <0.01 6.57 <0.001 Mean annual temp^2 0.69 2.05 <0.05 Mean annual temp^2 -0.18 -2.16 <0.05

Non-phylogenetic MAM of comparison A: extinction risk in frogs. d.f. = 506, R 2 = 0.69, MSE = 0.83

Coefficient t p-value Geographic range 1.11 6.24 <0.001 Geographic range^2 -0.17 -8.23 <0.001 Geographic range^3 <0.01 7.81 <0.001 Snout vent length 27.3 3.83 <0.001 Snout vent length^2 -6.92 -3.7 <0.001 Snout vent length ^3 0.58 3.6 <0.001 Mean annual temperature -8.24 -2.85 <0.01 Mean annual temperature^2 4.5 3.31 <0.01 Mean annual temperature^3 -0.74 -3.71 <0.001

138 Comparison B

Phylogenetic MAM of comparison B (RD species vs threatened species). n = 357, link = clog-log, deviance = 102.74, dfp (phylogenetic degrees of freedom) = 49.95, MSE = 0.29

coefficient t p-value log geographic range 0.20 1.92 0.06 Aquatic lifestage 0.75 3.06 0.00

Non-phylogenetic MAM of comparison B (RD species vs threatened species). n = 327, link = logit, df =326, MSE = 0.42

Coefficient z-value p-value log snout-vent length 2.57 3.13 0.002 log HPD -0.601 -2.694 0.007 Max temp -0.0176 -3.657 0.0003 Precip seasonality -0.0277 -3.665 0.00024 NPP -1.345 -2.234 0.025 log altitude -0.3355 -1.1912 0.055 log geographic range 0.257 1.914 0.056 aquatic lifestage 0.874 2.892 0.0038

139 Comparison C

Phylogenetic MAM of comparison C (enigmatic RD species vs threatened species). n = 120, link = clog-log, deviance = 30.02, dfp = 21.53, MSE = 0.26

Coefficient t p-value log eggs per clutch -0.75 -2.569 0.02 log altitude 0.31 1.77 0.09 0.04 aquatic lifestage 1.19 2.121

Non-phylogenetic MAM of comparison C (enigmatic RD species vs threatened species) n = 356, link = clog-log, df = 355, MSE = 0.37

Coefficient z-value p-value log geographic range -0.202 -2.298 0.0215 log HPD -0.599 -3.959 7.53E-05 Isothermality 0.0171 2.165 0.0304 Max temp -0.01179 -5.617 1.95E-08 Precip driest quarter 0.05499 2.728 0.00637 aquatic lifestage 0.9559 3.841 1.23E-04

140 Comparison D

Phylogenetic MAM of comparison D ( Bd + RD species vs all other Bd + species) n = 102, link = clog-log, deviance = 7.37, dfp = 16.94, MSE = 0.06

Coefficient t p-value log geographic range -1.57 -2.65 0.02 log altitude 1.15 2.66 0.02 aquatic lifestage 2.88 2.12 0.05

Non-phylogenetic MAM of comparison D ( Bd + RD species vs all other Bd + species). n = 102, link = clog-log, df = 101, MSE = 0.29

Coefficient z-value p-value log geographic range -1.723 -3.843 0.000122 log altitude 1.1753 3.312 0.000927 Max temp -0.023 -2.078 0.03773 aquatic lifestage 4.392 2.995 0.002747

141 Comparison E

Phylogenetic MAM of comparison E ( Bd + RD species vs Bd - RD species). n = 227, link = logit, deviance = 32.83, dfp = 33.82, MSE = 0.14

Coefficient t p-value log geographic range 0.47 3.24 0.003 actual evapotranspiration -0.02 -3.001 0.0052

Non-phylogenetic MAM of comparison E ( Bd + RD species vs Bd - RD species). n = 90, link = clog-log, df = 89, MSE = 0.40

Coefficient z-value p-value log clutch size 0.7136 2.811 0.00494 log HPD -1.1526 -3.995 6.48E-05 log altitude 0.4364 2.356 0.01849

142 Appendix 4.2: Mammalian regression models.

Comparison F

Phylogenetic MAM of comparison F (Extinction risk in Mammalia). d.f. = 302, R 2 = 0.273, MSE = 0.14

t p-value weaning age 2.26 <0.05 home range size 4.02 <0.001 population density -1.83 <0.1 population density^2 2.17 <0.05 geographic range size -3.23 <0.01 HPD 4.35 <0.001 HPD^2 -3.9 <0.001 geographic range size:HPD -3.39 <0.001 population density:HPD -2.84 <0.01

Comparison F

Non-phylogenetic MAM of comparison F (Extinction risk in Mammalia). d.f. = 254, R 2 = 0.4456 , MSE = 0.40

t p-value terrestriality -1.986 0.048 log habitat breadth 4.535 <0.001 log geographic range -7.781 <0.001 log litters per year -2.101 0.037 log neonatal body mass 4.75 <0.001 log home range 1 -2.361 0.019 log mean HPD 2.02 0.044 ETI 3.788 <0.001

143 Comparison G

Phylogenetic MAM of comparison G (Extinction risk in Chiroptera). d.f. = 368, R 2 = 0.419, MSE = 0.12

t p-value adult mass -1.87 <0.01 adult mass^2 2.19 <0.05 geographic range size 0.55 geographic range size^2 -2.86 <0.01 latitude 2.49 <0.05 HPD 5th percentile -1.52 HPD 5th percentile^2 2.75 <0.01

Comparison G

Non-phylogenetic MAM of comparison G (Extinction risk in Chiroptera). d.f. = 673, R 2 = 0.246, MSE = 0.42

t p-value log mean HPD -8.713 <0.001 log HPD 5th percentile 13.21 <0.001 ETI 0.224 0.099

144 Comparison H

Phylogenetic MAM of comparison H (Extinction risk in Marsupiala). d.f. = 145, R 2 = 0.649, MSE = 0.01

t p-value adult mass -4.13 <0.001 geographic range -15.57 <0.001 geographic range:adult mass 5.39 <0.001

Comparison H

Non-phylogenetic MAM of comparison H (Extinction risk in Marsupiala). d.f. = 62, R 2 = 0.55, MSE = 0.22

t p-value terrestriality 2.971 0.004 log geographic range -7.021 <0.001 log population density -4.447 <0.001 log age at sexual maturity -3.313 0.00154 mean temperature 2.883 0.0054 mean AET -3.4 0.00118 log mean HPD -2.835 0.00618

145 Comparison I

Phylogenetic MAM of comparison I (Extinction risk in Primates). d.f. = 129, R 2 = 0.346, MSE = 0.17

t p-value adult mass 0.74 geographic range size -5.67 <0.001 precipitation 3.05 <0.01 latitude -2.19 <0.05 HPD 5th percentile -1.53 HPD 5th percentile ^2 -2.1 <0.05 adult mass:latitude 2.86 <0.01

Comparison I

Non-phylogenetic MAM of comparison I (Extinction risk in Primates). d.f. = 86, R 2 = 0.5511, MSE = 0.66

t p-value log geographic range -7.755 <0.001 log home range 4.775 <0.001 log mean HPD 4.693 <0.001

146 Comparison J

Phylogenetic MAM of comparison J (Extinction risk in Rodentia). d.f. = 590, R 2 = 0.156, MSE = 0.09

t p-value geographic range size -7.25 <0.001 latitude 3.82 <0.001 HPD 5th percentile 2.52 <0.05

Comparison J

Non-phylogenetic MAM of comparison J (Extinction risk in Rodentia). d.f. = 287, R 2 = 0.256, MSE = 0.32

t p-value log geographic range -7.253 <0.001 log sexual maturity 4.576 <0.001 rate of HPD -4.326 <0.001 ETI 4.501 <0.001

147 Appendix 4.3: Optimal decision tree models.

Decision tree of comparison A (extinction risk in frogs) n =125, mean squared error = 0.7616

Geographic range < 21331km 2 |

3.4720 0.2679

148 Decision tree of Comparison B (RD species vs threatened species). n = 112, MSE = 0.217

Aquatic lifestage < 0.5 |

Net Primary Productivity < 1.075 0.4545

0.8765 0.3333

149 Decision tree of Comparison C (enigmatic RD species vs threatened species) n = 327, MSE = 0.228

Mean temp of warmest month< 21.4C |

Isothermality < 79.5 0.9286

Human population density < 2.32 0.5769

0.5349 0.1737

150 Decision tree of Comparison D ( Bd + RD species vs all other Bd + species) n = 101, MSE = 0.105.

Geographic range < 81.04km 2 |

Altitude < 288.59m 0.08219

0.14290 0.95240

151 Decision tree of Comparison E. (Bd + RD species vs Bd - RD species) n = 84,

MSE = 0.22.

Eggs per clutch < 10.64 |

HPD < 2.32 Altitude < 164.84

0.28570 0.86670 0.57140 0.09091

152 Decision tree of Comparison F (extinction risk in Mammalia) n = 400, MSE = 0.237.

Neonatal body mass < 937.3g |

Geographic range < 62498km 2 Litters per year <0.67

Rainfall < 27.26 Island.status < 0.5 3.0 1.65 Neonatal bm < 6447g 2.5 1.2 Neonatal bm < 46.92g Pop dens < 0.09 ETI < 0.32 0.33 2.0

1.4 0.08 0.22 1.5

153 Decision tree of Comparison G (extinction risk in Chiroptera) n = 268, MSE = 0.2007

Geographic range < 299479km 2 |

Body mass < 21.18g Latitude < 21.4013

Geographic range Precipitation < 176133km 2 < 64.4837 Actual evapotranspiration < 745.821 2.67 1.60 0.40 aetmean < 1374.02 5th % HPD Geographic range 0.89 < 6.49km -2 < 1149916km 2 0.06 0.60

ETI < 0.55 2.00 1.20 1.13 ETI < 0.41 0.095 0.36 1.16

154 Decision tree of Comparison H (extinction risk in Marsupiala) n = 48, MSE = 0.1316

Geographic range < 59749km 2 |

E.T.I. < 0.60 Interbirth interval < 152 2.25 Trophic level < 1.5 0.00 0.00 0.66 0.14

155 Decision tree of Comparison I (extinction risk in Primates) n = 98, MSE = 0.27027

Geographic range < 595169km 2 |

Body mass < 6435g Body mass < 3645g Actual evapotranspiration < 1142.06 0.07 0.38 1.50 Geographic range < 69251km 2 Geographic range < 70566km 2 Popn. Gestation HPD 60km 2 density < 236km -2 < 165 days 3.20 Latitude < 11.9 1.20 Popn density < 9.58km -2 2.00 0.29 1.20 3.00 2.00 0.60 2.20

156 Decision tree of Comparison J (extinction risk in Rodentia) n = 1213, MSE = 0.2779

2 Geographic range| < 39493km

Body mass < 455.64g HPD < 1.25km -2

HPD < 104km 2 HPD < 1.52km -2 Precipitation < 85.9673 Geographic range < 237875km 2

AET < 541.756 Body mass <12.7g Body mass < 13.6g 1.50 0.31 0.11 1.88 0.58 1.28 0.38 1.42 0.29 2.40 0.40

157 Appendix 4.4: Correlations coefficients of fitted values obtained from three different modelling techniques for ten different comparisons. df=degrees of freedom,

*p<0.05, **p<0.01, ***p<0.001,

PCM vs. TIPS PCM vs. DT TIPS vs. DT Comparison A 0.96*** 0.78*** 0.84*** (t=44.32, df =146) (t=15.39, df =146) (t=19.43, df =146) Comparison B 0.58*** 0.66*** 0.53*** (t=7.40, df =110) (t=9.09, df =110) (t= 6.51, df =110) Comparison C 0.40*** 0.20* 0.77*** (t=4.75, df =117) (t=2.14, df =111) (t=21.92, df =325) Comparison D 0.97*** 0.87*** 0.86*** (t= 38.66, df =100) (t=15.58, df =81) (t=14.92, df =81) Comparison E 0.48*** 0.43*** 0.71*** (t=5.14, df =88) (t=4.41, df =82) (t=9.30, df =82) Comparison F 0.81*** 0.66*** 0.70*** (t=20.29, df =215) (t=15.08, df =294) (t=15.06, df =231) Comparison G 0.72*** 0.55*** 0.35*** (t=26.98, df =663) (t=10.64, df =266) (t=6.02, df =264) Comparison H 0.67*** 0.63*** 0.75*** (t=7.46, df =69) (t=5.56, df =46) (t=6.53, df =34) Comparison I 0.90*** 0.75*** 0.75*** (t=19.50, df =87) (t=11.27, df =96) (t=9.19, df =65) Comparison J 0.71*** 0.39*** 0.45*** (t=16.97, df =290) (t=14.91,df =1208) (t=8.48, df =278)

158 Appendix 4.5: Mean squared error of extinction risk models obtained from different modelling techniques for ten different clades. Significant differences in the residual values, tested by use of Kruskal-Wallis with modelling technique coded as a factor and non-parametric multiple comparison tests, are represented by allocation of A, B and C. PCM TIPS DT

Comparison A*** 0.17 (A) 0.83 (C) 0.76 (B) df 2,836 , F 15.07

Comparison B 0.29 (B) 0.42 (C) 0.22 (A) df 2,793 , F 7.19

Comparison C 0.26 (B) 0.37 (C) 0.23 (A) df 2,800 , F 22.54

Comparison D 0.06 (A) 0.29 (B) 0.11 (A) df 2,284 , F 9.46

Comparison E 0.14 (A) 0.40 (B) 0.22 (A) df 2,398 , F 97.81

Comparison F 0.12 (A) 0.42 (B) 0.24 (A) df 2,1024 , F 19.85

Comparison G 0.20 (A) 0.42 (C) 0.21 (B) df 2,1286 , F 18.45

Comparison H 0.01 (A) 0.22 (B) 0.13 (A) df 2,303 , F 34.28

Comparison I 0.17 (A) 0.66 (B) 0.27 (A) df 2,364 , F 25.88

Comparison J 0.39 (B) 0.32 (B) 0.28 (A) df 2,3147 F 3.2

159 Appendix 4.6: Results of analysis of model residuals for phylogenetic signal. df = degrees of freedom, ?0.05-0.1,*<0.05, **<0.01, ***<0.001

Comparison TIPS DT PCM ______

A F=2.47***, df 31,481 F= 1.45, df 22,125 F= 5.03***,,df 31,481

B F= 4.26***, df 21,305 F= 3.39***, df 16,95 F= 20.43***, df 22,334

C F= 4.66***, df 22,333 F= 6.12***, df 21,305 F= 6.84***, df 17,102

D F= 0.51, df 13,88 F= 1.396, df 12,70 F= 1.52, df 13,88

E F= 0.58, df 15,74 F= 1.05, df 15,68 F= 2.25***, df 18,208

F F= 1.55**, df 70,193 F= 1.89***, df 72,327 F= 2.52***, df 71,299

? ? G F= 1.56 , df 14,662 F= 1.19, 13,254 F= 1.59 , df 14,747

H F= 1.209, df 13,57 F= 1.52, 12,35 F= 3.53***, df 17,173

? I F= 1.65 , df 12,77 F= 1.85, 12,85 F= 3.38***, df 12,168

? J F= 2.00 , df 23,268 F= 2.99***, 27,1185 F= 3.47***, df 27,1617

160 Chapter 5: Ground-testing a predictive model of Bd - susceptibility with the aim of directing further research.

Abstract

Predictive models of extinction risk and decline, such as those produced and assessed in chapters three and four of this thesis highlight species that, according to their biology and environment, are highly susceptible to extinction risk or population decline. The approach of identifying susceptible species may be particularly useful in prioritising conservation actions when management options are species-specific and labour intensive, as may be the case with amphibian declines caused by chytridiomycosis. In this chapter I combine the predictions made by a model of Bd susceptibility produced in chapter three with information on species’ population declines or the occurrence of mass mortality events to select four target species that had not previously been recorded as having been affected by Bd. I then investigate the possibility that chytridiomycosis may have been related to the decline or mortalities of those species. I find that sympatric species of one of the four species, Hyperolius cystocandicans , have been infected with Bd in the

Aberdares National Park, Kenya. The other three species, and sympatric species in the historic range did not appear to have been infected. I interpret these results and their conservation implications.

161 Introduction

Predictive models provide a useful tool in the field of conservation biology, and, as a consequence, the frequency of their use in ecological and conservation research has increased in recent years (Elith et al. 2006; Koh et al. 2004; Kotiaho et al. 2005). Modelling past trends and extrapolating the observed patterns taxonomically (Koh et al. 2004), geographically (Bielby et al. 2008; Cardillo et al. 2006) or temporally (Araujo et al. 2005a), may be useful in predicting future events (Sutherland 2006), or the status of regions or taxa outside of the original dataset. The predictions made may be used to direct future research, management or policy. In the field of conservation biology, predictive models have been used to address subjects such as extinction risk (Cardillo et al. 2006), species’ actual and realised distributions

(Elith et al. 2006), climatic change (Araujo et al. 2006), and species decline

(Sodhi et al. 2008).

Amphibian population declines are a subject that can greatly benefit from the use of predictive models as, once underway, some declines, particularly those occurring in connection with the disease chytridiomycosis, may be very rapid, and may be very difficult to arrest. The disease has been implicated in severe population declines in many species worldwide, and has been responsible for at least one species extinction (Schloegel et al. 2006), possibly many more

(IUCN et al. 2004; La Marca et al. 2005). Many of the most high profile declines have occurred in protected areas with relatively undisturbed habitat

(Berger et al. 1998; Bosch et al. 2001; Lips 1998; Lips 1999; Rachowicz et al.

2006), implying that traditional conservation measures, such as habitat

162 protection, may be ineffective against the threat posed by Bd . More effective conservation measures may include enforcing policy to restrict the movement of amphibians for trade purposes (Fisher and Garner 2007), monitoring the extraction and transportation of water (Johnson and Speare 2003), and ensuring disinfection of equipment between sites (Australian Government

2006) to reduce the risk of spreading the pathogen, and perhaps establishing ex situ populations (http://www.amphibianark.org/ 2008). Regardless of whether the measures are applied proactively or reactively, they are likely to be labour-intensive, financially expensive and logistically difficult, rendering the wide-scale application of such management options impractical or unfeasible. There is therefore a need to highlight which species should be the recipients of further conservation research and attention.

The output of predictive models may provide the first step in helping to highlight which species or locations should be the recipients of such management and may also serve to highlight those that may have already suffered the ill-effects of Bd infection. In the case of the latter, susceptible species that have suffered an observed population decline or mass mortality may warrant further attention. While linking the presence of a pathogen to the decline of a species is not straightforward (e.g. Daszak et al. 2003; e.g.

Daszak et al. 2005), a close link between the appearance of Bd and the decline of a species would be consistent with the hypothesis that chytridiomycosis was involved. Ideally, the role of Bd in a specific decline would be investigated by comparing the population dynamics of both infected and non-infected populations of a given species (Rachowicz et al. 2006).

163 Unfortunately, due to the lack of suitable sites and species for which long-term population data and both infected and non-infected populations are available, such studies are a rarity in current research and theory (but see Bosch and

Martinez-Solano 2006; Bosch et al. 2001; Lips et al. 2006; Rachowicz et al.

2006). The majority of studies linking disease to decline have therefore generally made the link between disease and decline retrospectively using historical specimens and dead or dying individuals (Bosch et al. 2001;

Burrowes et al. 2004; Muths et al. 2003).

The first step in such a retrospective process linking Bd to past declines would be to determine whether the pathogen was present or absent at the sites of decline or mortality within those species (Lips et al. 2008). Any species that were found to harbour Bd infection could then be identified and highlighted for further investigation regarding the role of Bd in that species decline or mortality. Given the absence of detailed population trend data for most amphibian species, in the interim it would be useful to have a surrogate measure of population trend to identify which susceptible species may already be in decline. In this chapter, I use the Rapid Decline status of species, or the occurrence of mass-mortalities, to identify susceptible species that may have already suffered negative consequences of Bd infection.

I selected a small number of target species with the aim of investigating the role of Bd in their decline. With the null hypothesis that Bd was not present at locations within the species’ former range, and therefore could not have been related to the target species’ declines, I collected tissue samples from the

164 target or sympatric species, and used molecular diagnostic tests to determine whether Bd was present at those locations. In the presence of Bd , I could therefore reject the null hypothesis and highlight these species as warranting further investigation regarding the role of Bd in their decline.

Methods

Target species

We selected four target species according to three criteria. First, species must have a high fitted value from the model of Bd susceptibility produced in chapter three or be the subject of a conservation project which would be compromised by the presence of Bd . Second, species must have suffered a population decline or a recent mass mortality. I defined a population decline as a species that has suffered a Rapid Decline: that is it has moved up a Red

List category, according to the 2004 GAA. Third, there must have been no records of Bd infection in the species.

The following four target species were selected from the candidate list:

Bufo brauni

Predicted susceptibility to Bd related decline = 0.89

Bufo brauni occurs in the West and East Usambaras, the Ulugurus, and the

Udzungwa Mountains in eastern Tanzania, with a total area of occupancy of less than 5000km 2, at altitudes of 750-1800m above sea level (a.s.l.) (IUCN et al. 2004). As a result of its restricted distribution, the species is classified as

Endangered B1ab(iii,v). In recent years the montane and sub-montane forest zones on which B. brauni relies have been degraded; as the species does not

165 tolerate modified habitat it has declined in population size. However, the high elevation at which the species resides in conjunction with its restricted range and reliance on water for breeding, results in a fitted value of 0.89, suggesting that it would be highly susceptible to Bd induced decline if infected.

Bufo lemur

Predicted susceptibility to Bd related decline = 0.28

The Puerto Rican crested toad ( Bufo lemur ) is currently restricted to a single location, Guanica National Forest, in the south of Puerto Rico. Historically B. lemur was distributed throughout both the north and south regions of the island. However, in recent years populations at many locations throughout the island have disappeared and, since 1988, no individuals have been observed in the northern section of the species’ range (Johnson 1999). In addition to the disappearance of populations at locations throughout the island, the one remaining breeding site has declined in size from 900 mature individuals in 1980, to approximately 100 at present (IUCN et al. 2004). On account of its restricted range and recent decline in population number and size, B. lemur was categorised as Critically Endangered on multiple criteria:

A2a, B1ab(v)+2ab(v), and C2a(ii) (IUCN et al. 2004). Although habitat loss is believed to have been responsible for the decline of this species, given the presence of Bd in other locations on the island (Burrowes et al. 2004), chytridiomycosis may have been involved. B. lemur did not have a particularly high susceptibility of decline (0.28). However, the species is the subject of a planned re-introduction which could be highly compromised if Bd

166 were implicated in the species’ decline. Thus information on the Bd infection status of amphibians at the site of reintroduction is very important.

Hyperolius cystocandicans

Predicted susceptibility to Bd related decline = 0.87

Hyperolius cystocandicans is endemic to the Kenyan highlands, occurring in less than 10 localities (IUCN 2004), and having a distribution of less than

20,000km 2. The restricted range and population decline of H. cystocandicans have resulted in the species being listed as Vulnerable under criterion B1 a,b(v) (IUCN et al. 2004). The GAA states that although its natural habitats

(montane grassland and pastureland habitats) are not seriously threatened in recent years there has been a general population decline for reasons that are not fully understood. The restricted range, reliance on temporary pools for breeding purposes, and high altitude (the type locality is at 2200m a.s.l.) resulted in a fitted value of 0.87 in my model.

Xenopus longipes

Predicted susceptibility to Bd related decline = 1.0

Xenopus longipes is a fully aquatic species that is endemic to Lake Oku, which has an area of less than 10km 2 at an altitude of 2200m a.s.l. in the

Cameroon highlands. Accordingly the species is listed as Critically

Endangered under criterion B1ab(v)+2ab(v) on account of its severely restricted distribution (IUCN et al. 2004). Several genera within the family

Pipidae ( Xenopus ; Weldon et al. 2004; Silurana; Parker et al.2002 ) and the

Xenopus genus ( Xenopus gilli , laevis , muelleri ; Weldon et al. 2004) are known

167 to be susceptible to Bd infection, while at least one member of the family is known to be susceptible to chytridiomycosis ( Silurana tropicalis ; Parker et al.

2002). On account of X. longipes’ high elevation, restricted range, and aquatic lifestyle, my model predicted its probability of decline on account of Bd infection as 1.0.

Collection of samples

I used samples of amphibian tissue from within the historical range of the four species to determine whether Bd was present, and therefore could have been involved in the observed mortalities or declines of these species. Although efforts were made to collect samples from the focal species, given the declines of my target species and hence the low abundance at which they were generally found, sympatric species which may act as reservoir hosts for

Bd (Daszak et al. 2003; Daszak et al. 2004) were also sampled. Tissue samples were obtained from a variety of sources as described below:

Type of tissue samples taken

For each species one or more of three types of tissue sample were collected: toe-clips, skin swabs using either fine-tip swabs (MW100, Medical Wire and

Equipment Ltd.) or interdental 3.2-6.0mm refill brushes (Oral-B Laboratories), or larval mouthparts. Single toe-clips were taken from adult animals and stored in 70% ethanol in individually marked tubes. Mouth-parts were taken from euthanized larvae and stored in 70% ethanol. Swab samples of live individuals were taken using fine-tip swabs by firmly running the swab over the ventral surface of the lower abdomen, drink-patch, all four limbs and all

168 digits of the individuals. Swab samples of archived specimens were taken by firmly running the dental refill brushes over the ventral surfaces of the lower abdomen, limbs and digits of each individual. All swab samples of both archived and live individuals were stored in individually marked tubes to avoid cross-contamination. In order to further minimise the possibility of cross- contamination of samples and subsequent false-positives, a new pair of disposable gloves were worn for each specimen handled, or in the case of archived specimens, for each specimen jar that was opened. Toe-clips, larval mouthparts and swabs of both live and dead specimens have all been successfully used in detecting Bd infection in the field and laboratory in previous studies (Hyatt et al. 2007; Kriger et al. 2006b; Soto-Azat 2007).

Bufo brauni,

Samples were collected from live individuals from the Amani Nature Reserve, in the east Usambara region of the Eastern Arc Mountains, Tanzania in a period of time between 27 th June and 20 th July 2006 (although precise GPS coordinates are unavailable for exact collection sites, the Amani Nature

Reserve is located between S5°14’10”-S5°04’30”, E38°30’34”-38°40’06”).

Historical specimens that had been preserved in ethanol in a time period between 1935 and 2005 were obtained from the archives of the Natural

History Museum, London.

Bufo lemur,

Collection and surveillance took place within a section of the historical range of the species in the northern karst region of Puerto Rico at El Tallonal private

169 land (N18°24’22.2”, E66°43’52.7”, 75m a.s.l.) in October 2005. El Tallonal was to be the site of reintroduction of larvae of the target species in 2006. All samples obtained were of sympatric species rather than B. lemur due to the extirpation of the target species in the northern part of its range.

Hyperolius cystocandicans,

Samples were collected from five sites within the historical range of the target species by Stefan Lötters in 2001 (Site 1: Aberdares National Park#1,

S00°31’29”, E35°22’35”, 2193m a.s.l.; Site 2:Aberdares National Park#2,

S00°24’41”, E36°43’27”, 3116m a.s.l.; Site 3:Aberdares National Park#3,

S00°28’51”, E36°44’56”, 2900m a.s.l.; Site 4:Chogoria Route, Mount Kenya,

S00°09’40”, E37°26’46”, 2997m a.s.l.; Site 5:Irangi Forest, Mount Kenya,

S00°20’50”, E37°29’02, 2013m a.s.l.). All tissue samples came from sympatric species rather than the H. cystocandicans itself because of the low abundance of the target species.

Xenopus longipes,

Tissue samples were collected from June-August 2006 in the north west province of Cameroon, from thirteen locations in the vicinity of the Kilum-Ijim

Forests that surround Mount Oku (N6°12'02.5", E10°27'35.4"; N6°12'11.9",

E10°22'42.5"; N6°12'43.8", E10°23'45.9"; N6°12'46.0", E10°23'39.3";

N6°12'50.7", E10°23'52.3"; N6°12'54.0", E10°24'24.7"; N6°12'55.0",

E10°23'56.4"; N6°13'04.1", E10°24'05.8"; N6°13'06.1", E10°24'06.8";

N6°13'39.6", E10°25'19.3"; N6°13'50.4", E10°24'53.7"; N6°13'56.9",

E10°25'14.0”; N6°17'3.6", E10°21'08.8"). During the sampling period the

170 collector, Thomas Doherty-Bone, observed unusual mortalities of X. longipes.

Samples of the target and sympatric species in the vicinity of Lake Oku were collected.

Molecular screening of samples

DNA was extracted from swab samples, toe-clips and larval mouthparts using a bead beating protocol (Boyle et al. 2004), extractions were diluted 1/10 with distilled water and then subjected to real time quantitative polymerase chain reaction (qPCR) developed specifically for the detection of Bd (Boyle et al.

2004). DNA extractions were performed on the tips of both fine-tip swabs and dental-brush swabs. All samples were screened in duplicate with four concentrations of standard Bd and a negative control. If the standard curves of the four concentrations of Bd did not amplify successfully then the samples in that plate were retested. The unit of infection burden was the Genomic

Equivalent (GE); one GE being equal to the amount of Bd -DNA present in the genome of a single zoospore.

Results

Bufo brauni

A summary of tissue samples collected from Amani Nature Reserve is presented in Table 5.1. In total, 267 swab samples were taken from 21 species, including 4 samples of live B. brauni , and 66 historical specimens.

None of the DNA extractions taken from the samples successfully amplified, suggesting that Bd DNA was not present in those samples. With 267 samples, there is a 95% probability that Bd would be detected in a population

171 of any size if the pathogen was present at a prevalence of 2% or higher

(Thrusfield 2005). For the table used to calculate the number of samples required to detect a disease at a stated prevalence with a confidence of 95%, please see appendix 5.1.

Bufo lemur

A summary of the tissue samples obtained from El Tallonal is presented in

Table 5.2. 179 samples were taken in total: 100 samples of larval mouthparts, and 79 swabs. None of the DNA extractions taken from the samples successfully amplified. With 179 samples, there is a 95% probability that Bd would be detected in a population of any size if the pathogen was present at a prevalence of 2% or higher (Thrusfield 2005).

Hyperolius cystocandicans

15 tissue samples of 10 sympatric species were obtained from the five sites surveyed (see Table 5.3). Of the fifteen samples, three successfully amplified with genomic equivalents of 234.8±116.1; 0.2+0.05; and 44.9±20.9. All three infected specimens, which were Ptychadena spp., Unknown spp. , and Amieta angolensis , came from the Aberdares National Park, although from different sites within the park. With 15 samples we can state with 95% confidence that

Bd would be detected if the pathogen was present at a prevalence of 20% or higher (Thrusfield 2005).

172 Xenopus longipes

A total of 283 tissue samples taken from 28 species were collected at Lake

Oku, including 74 samples of Xenopus longipes (see Table 5.4). None of the tissue extractions successfully amplified, suggesting that Bd was not present at this site. With 283 samples we can state with 95% confidence that Bd would be detected in a population of any size if the pathogen was present at a prevalence of 2% or higher (Thrusfield 2005).

Discussion

The results of the model ground-testing presented here may be used to draw preliminary conclusions about the threat mechanisms responsible for the declines of the four target species. In doing so, they may be helpful in directing more detailed research into the observed declines and perhaps the management for those species. On the basis of my results, I could reject the null hypothesis for only one of the four selected target species: H. cystocandicans . The infected status of this target species indicates that Bd is present in the surveyed populations, and that chytridiomycosis may be implicated in the declines of H. cystocandicans . In contrast, the non-infected status of the remaining three species provides some preliminary evidence that chytridiomycosis was not implicated in the declines of these species at the sites surveyed. I discuss the implications of these findings.

The fact that sympatric species of H. cystocandicans tested positive for Bd at three sites within the Aberdares National Park leaves open the possibility that

Bd was involved in the decline of this target species. Coupled with the

173 mysterious nature of the decline, which had no obvious causal mechanism associated with it (IUCN et al. 2004), the presence of Bd warrants urgent investigation into the role of chytridiomycosis in the decline of this target species. The presence of Bd within the Aberdares National Park suggests that population monitoring and further surveillance of Bd in the region are warranted to gain a better understanding of the threat posed by the pathogen in the Kenyan highlands, an area high in amphibian endemicity and species richness (BirdLife International 2007; Conservation International 2005).

Investigating the role of chytridiomycosis in declines of African species such as H. cystocandicans is further complicated by the hypothesis that Bd may have originated in sub-Saharan Africa, and may have a wide distribution, being endemic in some locations on the continent (Morehouse et al. 2003;

Weldon et al. 2004). Species from areas in which Bd is endemic may not be expected to suffer such serious consequences of infection as novel host species in other regions (Weldon et al. 2004), because of host-pathogen co- evolution resulting in the historical selection for genetically resistant animals

(Thrusfield 2005). In this respect, the co-evolution of Bd and host species may potentially bias the model predictions regarding species susceptibility.

However, despite the possible endemism of Bd ’ in parts of Africa, the exact location of origin is still not known, and the large land mass of the continent suggests that not all species have been historically exposed to Bd . Declines of other high elevation species within sub-Saharan Africa have been linked to chytridiomycosis (e.g. Nectophrynoides asperginis ; Weldon and du Preez,

2004), suggesting that some African species can be adversely affected by Bd .

174 For the remaining three target species I could not reject the null hypothesis that Bd was not present at the location of decline. The apparent lack of infected individuals at these sites has different conservation implications for each of the three species. B. lemur was to be the subject of a reintroduction at a site in the northern section of the species’ historical range. The fact that the location appears to be free from Bd suggests that this factor, at least, should not hinder the efforts to reintroduce the species at this location.

However, Bd has a confirmed presence on the island, and has been involved in the decline and possible extinction of at least three Puerto Rican species

(Eleutherodactylus eneidae , E. karlschmidti , E. jasperi ; Burrowes et al. 2004;

IUCN 2004). Further, the island has a number of species, both native and introduced, that may have acted as vectors of Bd in other regions (e.g.

Eleutherodactylus coqui , Bread and O’Neill 2005; Rana catesbeiana , Daszak et al. 2005, Garner et al. 2006). Therefore surveillance to monitor the spread of Bd in conjunction with threat abatement plans to prevent the transportation of the pathogen (e.g. Australian Government 2006) could be important to reduce further negative impacts of Bd in Puerto Rico. However, the apparent absence of Bd in the northern section of its range implies that the cause of the historical decline and extirpation of B. lemur is still at large and may therefore hinder the success of the reintroduction. Historically the drivers of a species’ decline have not always been taken into account in reintroduction and translocation efforts (Dodds and Seigel 1991). Habitat loss and degradation are thought to be the main mechanisms behind the historical decline of B. lemur (IUCN 2004), in which case, habitat protection of the remaining sites and successful reintroductions may aid recovery. However, Puerto Rico has

175 been the subject of a number of introductions of species such as Bufo marinus , and the Indian mongoose ( Herpestes auropunctatus ) which may adversely affect native species through competition or predation (e.g. Vilella and Swank 1993). These factors warrant further investigation as drivers of decline if the reintroduction is to be effective.

Although mass mortalities of Xenopus longipes were observed, and tissue samples were taken from dead and moribund individuals, no infected individuals were identified. On one hand, the apparent absence, or at least very low prevalence of occurrence, of Bd is a positive finding for the amphibians of the Cameroon highlands: Bd , a major threat to amphibian biodiversity, may not be present in the region, which harbours a number of endemic and restricted range species that may be naïve to the effects of introduced pathogens. On the other hand, the cause of the mortalities in X. longipes remains unknown and could be a potential threat to the amphibian assemblage of the region. The DNA extractions obtained from the tissue samples of X. longipes successfully amplified when screened for the major capsid protein of frog virus 3 (Duffus pers. comm.) indicating the potential presence of a ranavirus in those individuals. Ranaviruses have been implicated in mass mortality events in a number of species and locations

(Cunningham et al. 2007; Fox et al. 2006; Green et al. 2002). My results suggest that further efforts should be made to discover the cause of the Lake

Oku mortalities with particular attention to the role of ranavirus, and schemes to monitor the populations of susceptible species should be implemented.

The establishment of ex situ populations should also be a priority for a

176 species such as X. longipes , which is Critically Endangered, endemic, and has suffered an unusual mortality event.

There was no sign of Bd infection in the historical or recent specimens of B. brauni . The apparent lack of infection in this species suggests that Bd was not involved in the recent decline of the species. Other factors are therefore likely responsible for the increase in risk of extinction that this species has experienced in recent years. Habitat conversion of its forest habitat for farming, mining, and timber products is the most likely cause of the declines of this species, however, the presence Amani Nature Reserve and the

Udzungwa National Park within the species’ distribution provides some protection for the habitat of the species. Given the susceptibility of B. brauni and the presence of Bd in other regions in East Africa (Goldberg et al. 2007), including the parts of the Udzungwa mountains (Moyer and Weldon 2006), population monitoring and further surveillance of the distribution of Bd are recommended to assess future threats to the status of the species.

Despite the utility of the targeted sampling advocated here, there are limitations to using the approach in directing further conservation research.

Some of the problems stem from the difficulty in interpreting the results of a

Bd screen. In three of the species investigated here I could not reject the null hypothesis that Bd was not present at the site of decline. However, although we can say with 95% confidence that Bd is not present at a prevalence of above a stated threshold, we cannot state that a pathogen is entirely absent from a location, highlighting the difficulty in rejecting the possibility that Bd

177 was present. Additionally in calculating the required sample size we must make some assumptions about the diagnostic tests used, and the samples upon which our results are based. For the purposed of this analysis I have assumed that infection is equally probably across different epidemiological units (i.e. species and life-history stages within a species). Although Bd may well be transferable among species and life-stages (Rachowicz and

Vredenburg 2004), this assumption may be violated with some species and systems. A result of such violation would be the inaccurate calculation of the necessary sample size to detect infection at a stated population size and prevalence. To improve upon this, future efforts should use samples from the target species, and the same life-stage within a species whenever this is feasible. Further, the calculation of the required sample size also relies on the accuracy of qPCR, the diagnostic test for Bd , being perfect, which it may not be (Boyle et al. 2004). In the case of qPCR accuracy being less than 100%, the sample size for detection at a stated prevalence may be higher than calculated.

Even if the assumptions regarding the accuracy of qPCR and the infection among epidemiological units were not violated, the low prevalence at which

Bd may exist could result in inaccurate conclusions regarding the infection status of a site. According to the epidemiological model used in this study

(Thrusfield 2005), the sample sizes for the three target species from apparently uninfected sites were insufficient to detect Bd below a threshold of

2%. Previous studies have reported prevalences of less than 1% (Lips 1998;

Ouellet et al. 2005) which suggests that Bd may persist at such low levels.

178 Additionally, seasonal variation in the infection prevalence and presumably in infection burden and the zoospore density in the environment (Kriger and

Hero 2007) may lead to non-representative results in surveillance of Bd depending upon the time-period in which sampling is conducted. However, in larval oral-discs, toe-clips and swabs, we used the most accurate and cost- effective methods to sample for the presence of Bd , and our sample sizes were well above those required to detect Bd with 95% confidence at the stated prevalence and population size. The molecular diagnostic test (qPCR) used in this study is highly sensitive, being able to detect and quantify a single zoospore in a diagnostic sample (Boyle et al 2004), and has been recommended as the standard method for Bd detection (Kriger et al. 2006a;

Kriger et al. 2006b). Therefore even at times when prevalence and zoospore density is low, very light infections may be identified.

Even a positive result to a Bd test, suggesting that a species or population is infected, cannot be easily interpreted. The presence of Bd does not necessarily mean that chytridiomycosis is responsible for any amphibian population declines in that location (e.g. Daszak et al. 2005). In all cases, care should therefore be taken to ensure that other possible causal mechanisms are properly investigated. In the context of Bd there are some important examples of studies in which alternative causes may have been implicated in declines in areas of Bd infection (Daszak et al. 2005; Davidson et al. 2007; Whitfield et al. 2007), suggesting either the involvement of cofactors or that despite Bd infection, chytridiomycosis was not responsible for those declines. However, given the severity of the possible effects of Bd

179 (La Marca et al. 2005; Lips et al. 2006), an approach in which we take the presence of Bd as a potential threat to amphibian diversity is surely to be recommended, particularly given the speed of the effects Bd may have on a species or assemblage (e.g. Lips et al. 2006; Rachowicz et al. 2006).

Despite its limitations the approach outlined here is useful in terms of an initial survey to determine the necessity of further work on Bd in the region of a given decline. Implementing in-depth Bd and population monitoring schemes does not come cheaply. Therefore the approach of identifying declining species that are susceptible to the effects of Bd , and targeting those species for initial surveys is a relatively quick and inexpensive way of choosing which species should be the recipients of more focussed attention.

180 Tables

Table 5.1: Summary of Bd surveillance results for Bufo brauni .

Species Sample type n +ve % GE Afrixalus fornasini Swab 2 0 0 n/a Arthroleptis affinis Swab 5 0 0 n/a Arthroleptis stenodactylus Swab 8 0 0 n/a Boulengerula boulengeri Swab 1 0 0 n/a Bufo brauni Swab 4 0 0 n/a Bufo brauni (archived specimens) Swab 66 0 0 n/a Callulina kreffti Swab 7 0 0 n/a Hoplophryne rogersi Swab 4 0 0 n/a Hyperolius mitchelli Swab 41 0 0 n/a Hyperolius puncticulatus Swab 46 0 0 n/a Hyperolius spinigularis Swab 4 0 0 n/a Hyperolius spp. Swab 3 0 0 n/a Hyperolius tuberilinguis Swab 4 0 0 n/a Hyperolius viridiflavus mariae morph 1 Swab 11 0 0 n/a Leptopelis flavomaculatus Swab 29 0 0 n/a Leptopelis vermiculatus Swab 5 0 0 n/a Nectophrynoides tornieri Swab 8 0 0 n/a Phrynobatrachus kreffti Swab 12 0 0 n/a Rana angolensis Swab 3 0 0 n/a Schoutedenella xenodactyloides Swab 2 0 0 n/a Schoutedenella spp Swab 1 0 0 n/a Scolecomorphus vittatus Swab 1 0 0 n/a Total 267 0 0 n/a

181 Table 5.2: Summary of Bd surveillance results for Bufo lemur at El Tallonal,

Puerto Rico.

Species Sample type n +ve % GE Bufo marinus (adults) Swab 28 0 0 n/a Bufo marinus (larvae) Mouthparts 100 0 0 n/a Eleutherodactylus antillensis Swab 11 0 0 n/a Eleutherdactylus coqui Swab 35 0 0 n/a Eleutherodactylus spp. Swab 1 0 0 n/a Leptodactylus albilabris Swab 1 0 0 n/a Rana catesbeiana Swab 1 0 0 n/a Total 179 0 0 n/a

182 Table 5.3: Summary of Bd surveillance results for Hyperolius cystocandicans in central Kenya.

Site Sample type Species n +ve % 1 - Aberndares NP1 Toe-clip Hyperolius glandicolor 2 0 0 1 - Aberndares NP1 Toe-clip Ptychadena spp (adult) 4 1 25 1 - Aberndares NP1 Toe-clip Xenopus borealis 1 0 0 2 - Aberdares NP2 Toe-clip Amietia wittei 1 0 0 2 - Aberdares NP2 Mouthparts Unknown spp (larvae) 1 1 100 3 - Aberdares NP3 Toe-clip Amiettia angolensis 1 1 100 4 - Chogoria Route, Mt Kenya Mouthparts Unknown spp (larvae) 1 0 0 5 - Irangi Forest, Mt Kenya Toe-clip Afrana angolensis 1 0 0 5 - Irangi Forest, Mt Kenya Mouthparts Hyperolius spp (larva) 1 0 0 5 - Irangi Forest, Mt Kenya Mouthparts Ptychadena spp (larva) 2 0 0 Total 15 3 0.2

183 Table 5.4: Summary of Bd surveillance results for Xenopus longipes at Lake

Oku, Cameroon.

Species Sample type n +ve % GE Afrixalus fulvovittatus Toe-clips 6 0 0 n/a Arthroleptis adelphus Toe-clips 3 0 0 n/a Artholeptis cf. poecilinotus Toe-clips 6 0 0 n/a Artholeptis poecilinotus Toe-clips 6 0 0 n/a Arthroleptis variabilis Toe-clips 2 0 0 n/a Astylosternus ranoides Toe-clips 2 0 0 n/a Astylosternus rheophilus Toe-clips 4 0 0 n/a Bufo maculatus Toe-clips 2 0 0 n/a Bufo regularis Toe-clips 1 0 0 n/a Cardioglossa oreas Toe-clips 5 0 0 n/a Crotaphatrema lamoteii Toe-clips 1 0 0 n/a Hyperolius hieroglyphicus Toe-clips 20 0 0 n/a Leptodactylodon bicolor Toe-clips 3 0 0 n/a Leptodactylodon perreti Toe-clips 3 0 0 n/a Leptopelis modestus Toe-clips 1 0 0 n/a Phrynobatrachus cf. natalensis Toe-clips 1 0 0 n/a Phrynobatrachus cf. steinadachneri Toe-clips 30 0 0 n/a Phrynobatrachus cf. werneri Toe-clips 25 0 0 n/a Phrynobatrachus sp . Toe-clips 4 0 0 n/a Phrynobatrachus steinadachneri Toe-clips 39 0 0 n/a Phrynobatrachus werneri Toe-clips 10 0 0 n/a Ptychadena bibrioni Toe-clips 1 0 0 n/a Trichobatrachus robustus Toe-clips 5 0 0 n/a Werneria bambutensis Toe-clips 4 0 0 n/a Xenopus amieti Toe-clips 5 0 0 n/a Xenopus laevis Toe-clips 1 0 0 n/a Xenopus longipes Toe-clips 74 0 0 n/a Xenopus sp. Toe-clips 2 0 0 n/a Unidentified frogs Toe-clips 17 0 0 n/a Total 283 0 0 n/a

184 Appendix

Appendix 5.1: Sample size required for detecting disease where the probability of finding at least one case in the sample is 5%.

Percentage of diseased animals in the population. Population size 50% 40% 30% 25% 20% 15% 10% 5% 2% 1% 0.5% 0.1% 10 4 5 6 7 8 10 10 10 10 10 10 10 20 4 6 7 9 10 12 16 19 20 20 20 20 30 4 6 8 9 11 14 19 26 30 30 30 30 40 5 6 8 10 12 15 21 31 40 40 40 40 50 5 6 8 10 12 16 22 35 48 50 50 50 60 5 6 8 10 12 16 23 38 55 60 60 60 70 5 6 8 10 13 17 24 40 62 70 70 70 80 5 6 8 10 13 17 24 42 68 79 80 80 90 5 6 8 10 13 17 25 43 73 87 90 90 100 5 6 9 10 13 17 25 45 78 96 100 100 120 5 6 9 10 13 18 26 47 86 111 120 120 140 5 6 9 11 13 18 26 48 92 124 139 140 160 5 6 9 11 13 18 27 49 97 136 157 160 180 5 6 9 11 13 18 27 50 101 146 174 180 200 5 6 9 11 13 18 27 51 105 155 190 200 250 5 6 9 11 14 18 27 53 112 175 228 250 300 5 6 9 11 14 18 28 54 117 189 260 300 350 5 6 9 11 14 18 28 54 121 201 287 350 400 5 6 9 11 14 19 28 55 124 211 311 400 450 5 6 9 11 14 19 28 55 127 218 331 450 500 5 6 9 11 14 19 28 56 129 225 349 500 600 5 6 9 11 14 19 28 56 132 235 379 597 700 5 6 9 11 14 19 28 57 134 243 402 691 800 5 6 9 11 14 19 28 57 136 249 421 782 900 5 6 9 11 14 19 28 57 137 254 437 868 1000 5 6 9 11 14 19 29 57 138 258 450 950 1200 5 6 9 11 14 19 29 57 140 264 471 1102 1400 5 6 9 11 14 19 29 58 141 269 487 1236 1600 5 6 9 11 14 19 29 58 142 272 499 1354 1800 5 6 9 11 14 19 29 58 143 275 509 1459 2000 5 6 9 11 14 19 29 58 143 277 517 1553 3000 5 6 9 11 14 19 29 58 145 284 542 1895 4000 5 6 9 11 14 19 29 58 146 268 556 2108 5000 5 6 9 11 14 19 29 59 147 290 564 2253 6000 5 6 9 11 14 19 29 59 147 291 569 2358 7000 5 6 9 11 14 19 29 59 147 292 573 2437 8000 5 6 9 11 14 19 29 59 147 293 576 2498 9000 5 6 9 11 14 19 29 59 147 294 579 2548 10000 5 6 9 11 14 19 29 59 147 294 581 2588

185 Chapter 6: Distribution of Batrachochytrium dendrobatidis on the Mediterranean island of Sardinia.

Abstract

Batrachochytrium dendrobatidis ( Bd ), the causative agent of the amphibian disease chytridiomycosis, has been implicated in the decline of amphibians around the world, and has been reported on every continent on which amphibians live. Although Bd has a wide but patchy distribution within

Europe, clinical signs of chytridiomycosis had been limited to Peñalara

National Park, Spain. However, a second occurrence of chytridiomycosis was recently reported in Euproctus platycephalus from the Mediterranean island

Sardinia. Given the endemicity and threat level of species on the island, the occurrence of chytridiomycosis is a serious cause for concern for the conservation of the amphibians of Sardinia. In order to gain a better understanding of the distribution of Bd on Sardinia, I conducted an opportunistic surveillance for the pathogen on the island sampling 528 individuals of five species at 19 sites. Bd was found to be present in two locations in two separate drainage basins in the vicinity of the northern

Limbara mountains. Further, both sites are known to have experienced unusual mortalities of the Tyrennhian painted frog, Discoglossus sardus in the recent past. In addition to the first described occurrence in the Sette Frattelli mountains, Bd is now known to be present in at least two additional populations of amphibians in Sardinia. I discuss the distribution of Bd on the

186 island the implications of these findings for the conservation of Sardinian species.

Introduction

The emergence and spread of Batrachochytrium dendrobatidis (Bd), the causative agent of chytridiomycosis, has been strongly linked to the global decline of amphibian species. Chytridiomycosis has been implicated in mass mortality events (Bosch and Martinez-Solano 2006), population declines

(Bosch et al. 2001; Lips et al. 2006), and both regional and global extinctions

(La Marca et al. 2005; Schloegel et al. 2006) since the mid-1980s. As a result of the severity of its effects, chytridiomycosis has been described as “the worst infectious disease ever recorded amongst vertebrates in terms of the number of species impacted, and its propensity to drive them to extinction”

(ACAP 2005). So far, Bd is known to occur on six continents and to have infected over 450 amphibian species

(http://www.parcplace.org/bdmap2008update.html ) , although the real figure is probably much higher due to current biases and limitations in sampling.

Given the potential severity and rapidity of its effects, gaining a better understanding of the epidemiology and distribution of the pathogen is urgent and of global importance.

Within Europe, Bd is known to have a relatively wide, although patchy, distribution (Garner et al. 2006; Garner et al. 2005), and to have infected at least 27 of the 81 European amphibian species (Bosch and Martinez-Solano

2006; Bosch et al. 2001; Bovero et al. in press; Fisher and Garner 2007;

187 Garner et al. 2006; Garner et al. 2005; Simoncelli et al. 2005; Stagni et al.

2002; unpublished data) . However, despite our improving knowledge of the geographic and taxonomic distribution of Bd within Europe our knowledge of host-pathogen dynamics and its effects on European species remains relatively poor. The best-studied system in Europe is Peñalara National Park in central Spain, which until recently was the only European location where the occurrence of chytridiomycosis had been observed (Bosch et al. 2001).

The negative effects of chytridiomycosis at Peñalara were first noticed in the common midwife toad, Alytes obstetricans , between 1997 and 1999, when mass mortalities led to population declines and the extirpation of the species in 86% of the ponds in the natural park (Bosch et al. 2001). Since then, chytridiomycosis has been implicated in mass mortalities of the fire salamander ( Salamandra salamandra) and the common toad ( Bufo bufo) in

Peñalara, with population declines of the former species also being recorded

(Bosch and Martinez-Solano 2006).

Recently, a second occurrence of chytridiomycosis in European amphibians was reported in the endemic Sardinian newt, Euproctus platycephalus

(Bovero et al. in press) . In 2005, several individuals were observed with possible signs of chytridiomycosis at a site in the south of Sardinia, including ulcerative damage and degradation to the digits of both fore- and hindlimbs.

Three of the five individuals exhibiting these signs were subsequently found to be infected with Bd (Bovero et al. in press) . Although ulceration may be associated with other amphibian diseases (e.g. Ranavirus; Cunningham et al.

1996), and damaged limbs have been observed in other aquatic salamanders

188 infected by Bd (e.g. Cryptobranchus alleganiensis ; Missouri Department of

Conservation 2007), at present it is not clear whether ulceration and digit loss is a reliable sign of Bd infection.

E. platycephalus is a highly specialised, fully aquatic newt (Andreone and

Luiselli 2000; Van Rooy and Stumpel 1995), that inhabits still or running freshwaters, including rivers, streams and pools, usually in hilly or mountainous areas between 500-900m above sea level (asl) (Lecis and

Norris 2003). The historical range of E. platycephalus covered most of the mountainous areas of the island (Carruccio 1869; Gene 1839). However, the species is currently restricted to mountain regions on the east side of Sardinia in the northern Limbara region, the central Gennargentu region and southern

Sette Fratelli region (Bovero et al. 2005; Lecis and Norris 2003). Genetic studies suggest that the E. platycephalus populations of the three mountain regions are genetically distinct, and should therefore be considered as separate management units (MU) (Lecis and Norris 2004).

Declines of E. platycephalus in all three MUs have been reported since the early 1980s in terms of both the number of populations and in mature individuals per population (Colomo 1999; Puddu et al. 1988) with presumably few viable populations remaining (Lecis and Norris 2003). The species has been described as Europe’s rarest amphibian (Van Rooy and Stumpel 1995) and “the nearest urodele to extinction in Europe” (Grossenbacher, cited in

Andreone and Luiselli 2000), and is an example of the urgent need for amphibian research in Europe (Pasmans et al. 2007). As a result of its

189 restricted range and the declines that it has experienced, the species is classified as Endangered according to the IUCN Red List (IUCN et al. 2004), and receives conservation attention at both the regional and national levels in

Italy. Degradation of habitat via water extraction, pollution, illegal fishing methods, and introduced fish species (Read 1998; Schenk et al. 1995) have all been proposed as causes of decline and local extirpation. The presence of

Bd on the island suggests that disease may be another, potentially important, driver of decline. However, at present we know little about the severity of the effects of Bd infection on Sardinian amphibians.

The presence of Bd on the island, and chytridiomycosis in E. platycephalus

(Bovero et al. in press) provide cause for serious concern. Of Sardinia’s eight native amphibian species, five are endemic, and a further two have extremely restricted ranges limited to Sardinia, Corsica and nearby islands. Sardinia’s amphibian fauna includes five species of the genus Speleomantes , which represent over half of the eight species of plethodontid salamander found outside of the Americas; one of only six extant members of the genus

Discoglossus ; and one of only two currently recognized members of the genus Euproctus. Of these species, one is near threatened ( Speleomantes imperialis ), two are vulnerable ( S. genei and S. flavus ), and two are endangered ( S. supramontes and E. platycephalus ) according to the IUCN

Red List. Given the high endemicity and level of extinction risk of Sardinian species, identifying which species are susceptible to infection and may act as alternative hosts for Bd is of great importance to conservation efforts for

Sardinian amphibians (Van Rooy and Stumpel 1995).

190

Here I describe a systematic surveillance of Bd on Sardinia, focussing on E. platycephalus . In so doing, I aim to improve our knowledge of the distribution of Bd on Sardinia and how the observed distribution fits in with current epidemiological theory. By conducting an opportunistic screening I also aim to determine whether alternative hosts exist among Sardinia’s amphibian species, and whether the ulceration and loss of digits in E. platycephalus is a reliable indication of Bd infection. I aim to use the information obtained to determine which species and sites should be the focus of conservation management plans such as the collection of individuals for ex situ conservation, and of special efforts to prevent the further spread of the pathogen on the island.

Methods

Study system

In each of our three collecting trips I concentrated on sampling a different MU, aiming to collect samples from at least six sites within each MU. Sites are defined here as stretch of river or stream no more than 100m in length and separated from other sites by at least 5km. The three collection trips ran from the 10 th -15 th May, 7 th -14 th August, and 1 st -10 th November 2007, and focussed upon the Sette Fratelli (southern MU), Limbara (northern MU), and

Gennargentu regions (central MU) respectively. Although species presence and abundance, and potentially also the presence of Bd (Kriger and Hero

2007), vary through the year, it was logistically impossible to conduct field- work in all regions simultaneously for this study. However, water temperatures obtained at each site during the collecting period were between

191 11-20.0°C which are all amenable for the survival and reproduction of the pathogen (Berger et al. 2004; Longcore et al. 1999; Piotrowski et al. 2004).

Sample sizes

At each site I aimed to collect and sample at least 30 individuals of E. platycephalus . A sample size of 30 allows the determination of whether Bd was present in a population of any size with 95% confidence, assuming a prevalence of Bd of over 10% (Thrusfield 2005). Although it would be desirable to use a lower prevalence threshold, the required sample size increases very rapidly to a point that would be unfeasible for E. platycephalus collection ( e.g. for a population with an assumed size of 100 individuals, 45 would need to be sampled for a population with a 5% prevalence, or 96 for a population with a prevalence of 1% c.f. 29 with a prevalence of 10%)

(Thrusfield 2005).

Diurnal visual surveys for E. platycephalus were conducted, searching each section of stream bed and each pool exhaustively by lifting rocks, substrate and leaf litter and capturing each animal encountered. In deeper pools, a scuba mask, snorkel, and long-handled net were used to capture any specimens that were present. Whenever possible, transects ran downstream to avoid the possibility of spreading chytrid upstream into non-infected areas.

Target Species

As infection and signs of chytridiomycosis on Sardinia had previously only been observed in E. platycephalus we concentrated our survey efforts on this

192 species. However, due to the potential importance of alternative hosts as reservoirs of Bd , I also sampled sympatric species if they met any of the following criteria:

- As introduced species have been suggest as a point of introduction and vectors of Bd (Fisher and Garner 2007; Garner et al. 2006) they are an important target for sampling on Sardinia. One population of introduced waterfrogs ( Pelophylax/Rana spp. hybridogenetic complex) was sampled by capturing frogs in the pond by hand.

- Species that have suffered unusual mortality events may have been affected by chytridiomycosis and therefore should be a high priority for Bd screening, particularly those populations in which mortality events have occurred. For example, in June 2004 six dead adult individuals of

Discoglossus sardus were discovered in a stream feeding a freshwater lake,

‘Laghetto dei Pompieri’, located in the Limbara Mountains in northern Sardinia

(N40° 50’ 20.0”, E9° 8’ 35.0”). (Bovero pers comm). Similarly, in May 2006, approximately 15-20 D. sardus mortalities were discovered within close proximity of one another in the water body at Monte Olia forest camp, also in the vicinity of the Limbara Mountains (N40 ° 44’ 45”, E9 ° 21’ 40”). D. sardus was therefore an important species to sample.

- Congeners of species that have been infected with Bd in other locations may be suitable alternative hosts for Bd. For example, Hyla arborea has been infected by Bd on continental Europe, implying that the closely related

H. sarda (formerly a sub-species of H. arborea ) is a possible alternative host.

I conducted a number of nocturnal visual surveys to sample both

Discoglossus sardus and Hyla sarda . The search process for D. sardus was

193 similar to that used for E. platycephalus , while for H. sarda calls were followed and vegetation was searched for individuals.

- Due to the endemicity and high level of extinction risk of the plethodontid salamanders of Sardinia, sampling populations of the genus Speleomantes was a priority to ascertain whether populations are threatened by chytridiomycosis. Therefore I additionally sampled a population of

Speleomantes imperialis from the Sette Fratelli region.

Collection and field hygiene

All amphibians were caught on sight and kept individually in re-sealable plastic bags to avoid cross infection. All individuals were sexed, weighed, measured (snout-vent length for anurans, both snout-vent length and total length for caudates), classified to life-history stage as well as surveyed for distinguishing features such as digit loss or presence of lesions or ulceration.

Sampling for infection consisted of taking digit-tips and/or swab samples from each individual that was captured. Digit-tips were collected from the first digit on the right rear foot and stored in 70% ethanol. Swab samples were taken by firmly running a fine-tip swab (MW100, Medical Wire and Equipment Ltd.) over the lower abdomen, drink-patch, all 4 limbs, and digits of all limbs. A clean pair of disposable powder-free nitrile gloves were worn from each individual handled. Dead individuals found at any site were collected and stored in 70% ethanol. Tissue samples of dead individuals were taken from hindlimb digits and the drink-patch for DNA extraction. Additionally, in order to visualise any Bd infection of the epidermis, tissue samples from any infected dead specimens collected were embedded in paraffin, stained,

194 sectioned and submitted to a commercial laboratory for histological processing.

Between sampling sites all equipment was disinfected for 10 minutes in household bleach containing at least 1% sodium hypochlorite (Australian

Government 2006; Johnson et al. 2003) and rinsed well with water.

Laboratory techniques

DNA was extracted from both the swab samples and tissue samples using a bead beating protocol (Boyle et al. 2004). Extractions were diluted 1/10 with distilled water and then subjected to real time quantitative polymerase chain reaction (qPCR) developed specifically for the detection of Bd (Boyle et al.

2004). All samples were screened in duplicate with four concentrations of standard Bd and a negative control. The unit of infection burden was the

Genomic Equivalent (GE); one GE being equal to the amount of Bd -DNA present in the genome of a single zoospore.

Results

Between May and November 2007, a total of 528 individuals at 18 sites were sampled in the three mountain regions of Sardinia (Fig. 6.1, Table 6.1). The predetermined level of thirty specimens was achieved at 10 of the 18 sites.

Region 1 - Sette Fratelli

A total of 228 amphibians were sampled at seven sites in the Sette Fratelli mountain range (see Figure 6.2 and Table 6.1). The specimens included 141

195 E. platycephalus , 61 Speleomantes imperialis , 16 D. sardus , and 10 H. sarda .

At site M1 (Rio Marani) tissue damage was observed on the digits of either front or hind limbs of a relatively large number of individuals (10 of 30 collected, perhaps indicating chytridiomycosis (e.g. see figure 6.3). Sufficient individuals were collected at four of the sites in Sette Fratelli (M1, M4, M5 and

M6) to have 95% confidence of detecting Bd assuming a prevalence of 10% .

In the remaining three sites enough samples were collected to have 95% confidence of Bd detection assuming a prevalence of 15% (n=19; Thrusfield

1995). Despite this MU having a site with a high prevalence of damaged digits (site M1; Rio Marani) and also including the site of first Bd description in

Sardinia (Bovero et al. in press ), no successful amplifications of Bd DNA were obtained from these samples.

Region 2 - Limbara

Between 7th and 14th of August in the northern Limbara mountain range more sites (n=8), and individual amphibians were sampled (n=267) than in the other regions (see Figure 6.4). The majority of samples were H. sarda

(n=154) , as the period of time in which sampling was conducted coincided with the metamorphosis and emergence of this species from natal ponds.

Additionally tissue samples were collected from an introduced population of water frogs ( Pelophylax/Rana spp. hybridogenetic complex) (n=32) to assess the possibility that this population acts as a reservoir or possible point of introduction for the fungus on the island (Fisher and Garner 2007). D. sardus was sampled at four sites (n=28). 62 samples of E. platycephalus were collected, including 53 individuals from one site outside of the documented

196 range of this species in Limbara (site A8). Of the eight sites sampled in the

Limbara region, six yielded enough samples to detect Bd assuming a prevalence of 10% (A2, A3, A4, A5, A6, A8), one (A7) had enough samples to detect Bd given a prevalence of 15% (required n=19), and at the remaining site (A1) enough samples were collected to detect Bd given a prevalence of

20% (required n=14).

DNA extractions from samples collected in the Limbara Mountains successfully amplified in three of 17 samples taken from apparently healthy metamorphic individuals of Discoglossus sardus sampled at site A7 (table

6.1). The mean GE for these amplifications were 0.11 ge +0.03, 10.81 ge

+0.60, and 83.18 ge +7.27. We obtained a single preserved adult specimen of undetermined sex from the mass mortality event at Monte Olia Forestry

Park in 2006 in the Limbara region (N40° 44’ 45”, E9° 21’ 40”) described in the methods section. Tissue extractions from the dead specimen successfully amplified, with a GE of 55.2+ 13.5 for the digit tissue and a mean GE of

78.3+15.7 for the drink patch tissue. Tissue samples from the specimen were processed and inspected as histological samples. Unfortunately very few epidermal cells remained after the histological preparation, most likely because the corpse was already badly decomposed when it was collected in

2006. Inspection of Figure 6.5 suggests that the two positive sites in the

Limbara region are in separate drainage basins.

Region 3 - Gennargentu

Between 1st and 10th November a small number of samples (n=33) were collected in the central region of Gennargentu, taken from three sites at which

197 E. platycephalus were found (see Figure 6.6). In order to increase the sample size for the Gennargentu region I included specimens of E. platycephalus collected in 2005/06 (n=87). No successful amplifications of DNA extractions were obtained from any of the samples collected in Gennargentu.

Discussion

These results, in conjunction with previous studies on Sardinia, describe the presence of Bd at a minimum of three sites: the Maidopis stream system in the southern Sette Fratelli region (Bovero et al. in press ), and two sites located in separate water basins in the vicinity of the Limbara Mountains in northern Sardinia (See Table 6.1 for details of all sites). Here we discuss the implications of our results focussing on the potential susceptibility of Sardinian species, the possible epidemiological reasons for the observed distribution, and future directions for research and management of Bd on Sardinia.

Increasing our knowledge of the emergence and epidemiology of Bd in a given location may help in our efforts to implement the most suitable and effective conservation management. The recent emergence of Bd has been explained by two mechanisms that are not mutually exclusive: the Novel

Pathogen Hypothesis (NPH) and the Endemic Pathogen Hypothesis (EPH)

(Rachowicz et al. 2005). The former suggests that Bd has recently spread into new geographic areas, while the latter states that hosts are becoming more susceptible to pre-existing infections as a result of changes to the environment. Determining which of these hypotheses is most accurate in a given location (and indeed more generally) will have important implications for

198 disease control: a novel origin would require identification and management of vectors of spread, whereas an endemic pathogen would require management of co-factors and synergistic agents. Our results suggest that Bd is present at only three of the surveyed sites (Table 6.1; Figure 6.1). In theory, either the

EPH or NPH could explain the observed distribution: the pathogen could have been recently introduced into those sites, or it could be endemic to a small number of populations on the island. However, the rate of spread of Bd in other infected regions has been estimated as 40-60 km/yr in Queensland,

Australia (Alexander and Eischeid 2001), 15-33km/yr in Costa Rica and

Panama and 25-282km/yr in the South American Andes (Lips et al. 2008).

With those estimated rates of spread, it seems likely that an endemic pathogen would have a wider distribution and infected a larger number of populations, given a suitable method of dispersal and suitable hosts and conditions at other sites. As Bd may potentially be transported via amphibians (Fisher and Garner 2007), water (Johnson and Speare 2003), substrate (Johnson and Speare 2005) and even bird feathers (Johnson and

Speare 2005), such vectors are present and available on Sardinia, as are apparently suitable habitat, conditions and hosts elsewhere on the island. It therefore seems likely that, if Bd were endemic it would be present at more sites than our results suggest supporting the hypothesis that Bd is a novel pathogen that was recently introduced to Sardinia. However, as no epidemiological analyses were conducted on our analyses it is not possible to disentangle whether the EPH or NPH is responsible for the observed distribution of Bd on Sardinia. With further surveillance, and consequent larger samples sizes, such epidemiological analyses may be possible and in

199 conjunction with genotyping of the Bd strains present, we may be better able to understand the distribution and epidemiology of the pathogen on Sardinia.

Further investigation of the suitability of presumed non-infected sites for Bd might also be warranted to determine whether conditions conducive to the pathogen prevail throughout the year. For example, does the complete drying of some streams during summer months preclude the possibility of Bd persisting at those sites?

In total five species of amphibians were screened for Bd infection on Sardinia

(table 6.1). In addition to E. platycephalus only D. sardus could be confirmed as an alternative host of Bd . The results of this screening cannot confirm that native species such H. sarda , or S. imperialis are acting as an alternative host for Bd on the island. Additionally, based on the evidence from this surveillance it does not appear that the population of introduced

Rana/Pelophylax species complex surveyed (site A4) have acted as a point of introduction or a reservoir for Bd on Sardinia. However, given the history of invasive species as potential routes of Bd introduction, identification of further populations and investigation of their Bd status may be warranted. The two infected populations of D. sardus identified in this study (A7 and Monte Olia) have both suffered unusual mortalities in recent years, meaning that all sites at which infection has been found on Sardinia have shown either signs of chytridiomycosis (Rio Maidopis; Bovero et al. In press ) or have experienced mass mortalities (e.g. Laghetto dei Pompieri in 2004 and Monte Olia 2006).

Additionally, both D. sardus and E. platycephalus have aquatic life-stages and

200 relatively restricted, high altitude distributions all of which may increase a species’ susceptibility to Bd related decline (see chapter 3).

In combination, their biological profiles and the signs of chytridiomycosis in infected populations suggest that Sardinian amphibian species such as E. platycephalus and D. sardus could face severe consequences as a result of infection.

Given the potentially severe effects of chytridiomycosis on Sardinian species, there is a need for careful direction of future work on the distribution and effects of Bd on the island. The results of the opportunistic surveillance described in this chapter can be used to set such aims and objectives for future work. A priority would be to increase the sample size of amphibians screened at each site in all regions of Sardinia in order to improve our ability to detect Bd at a lower prevalence. Further, false negative results could also occur as a result of PCR inhibitors in the environment. Future work should incorporate internal positive controls (IPC) in at least a subset of the samples from each site in order to establish whether inhibition has occurred as a result of environmental inhibitors (Walker et al 2007).

Ideally collections would be conducted in all regions simultaneously, when amphibian densities are high, and conditions are most likely to suit high Bd prevalence (Kriger and Hero 2007). From our experience early summer

(May) would be the most suitable time to collect samples. Later in the summer (August) many streams had dried out, whereas in winter (November) the water and air temperatures were too low to obtain enough amphibians.

201 Within the most suitable season, it may be useful to sample at different times during the day. Little is currently known about the activity cycle of the species, which therefore may be active and easier to find at night. The logistics of sampling stream-dwelling E. platycephalus at night made it impossible to do so during this study. However, with a better knowledge of the inhabited sites, efforts to conduct nocturnal collections of E. platycephalus would be a worthwhile endeavour. Finally, in the absence of a sufficient number of amphibians, environmental sampling (Walker et al. 2007) for Bd could be a useful tool in future efforts to map the distribution of Bd on

Sardinia.

The results of this surveillance can also be used to prioritise populations and regions for directing future efforts to map Bd on Sardinia at a finer scale. The first priority should be to screen the infected populations of D. sardus at Monte

Olia and Laghetto dei Pompieri and E. platycephalus at Maidopis more extensively in terms of both the number of sites screened in the surrounding area and the number of individuals sampled per site. As a future direction we recommend focussed efforts to sample and observe the population for clinical signs of disease and unusual mortality at the time of breeding and metamorphosis when densities of amphibians are highest, and the effects of

Bd most likely to be observable (Rachowicz et al. 2006). Based on the occurrence of abnormalities in the digits of E. platycephalus in apparently non-infected individuals at site M1 (Rio Marani), it appears that loss of digits is not a reliable sign of infection status. However, in other species of aquatic salamander (e.g. Cryptobranchus alleganiensis ; Missouri Department of

202 Conservation 2007) it appears that the presence of lesions and Bd infection may be linked, so perhaps this association warrants further investigation in E. platycephalus . The role of other potential cause of the loss of digits should also be examined. For example, ranaviruses have been linked to mass mortality events in a number of amphibian species (Cunningham et al. 2007;

Fox et al. 2006), and one of the signs of infection by the virus is the loss or deformation of digits. The incidence of ranavirus on Sardinia, particularly in populations where individuals have deformed limbs and digits, may be worth further investigation.

Islands such as Sardinia represent an opportunity to tackle conservation issues on a more manageable scale than continental sites facing similar problems. If, as our results suggest, Bd is limited to a relatively small number of sites on Sardinia, efforts must be made to prevent the spread of the pathogen, and to limit its impact at infected sites. Two obvious modes of dissemination of Bd , which may remain infective in water or substrate for up to seven weeks (Johnson and Speare 2003; Johnson and Speare 2005), are the extraction and transportation of water and the movement of human traffic. In addition to reducing the amount of available habitat and increasing disease transmission among individuals (Burrowes et al. 2004), the extraction and translocation of water from streams and ponds could also aid the spread of

Bd . Careful monitoring of the movement of extracted water on the island would reduce the risk of disease transmission via this potential route of pathogen spread. Additionally, human traffic, specifically forestry workers who may travel between multiple sites in one day, could transport Bd in

203 substrate or water attached to their equipment, vehicles and clothing

(Johnson and Speare 2005). Raising awareness of the potential risk of Bd transportation, and the ease with which forestry equipment and footwear can be disinfected with suitable anti-fungal agents between sites would be a simple way to reduce the likelihood of disease transmission via this route

(Australian Government 2006; Johnson et al. 2003).

To summarise, our results suggest that, in addition to the multiple threats already faced by Sardinian amphibians, chytridiomycosis poses a further risk; they also provide us with future directions for sampling efforts. In the context of chytridiomycosis, a better understanding of the distribution and epidemiology of the pathogen would aid policy-makers and conservation practitioners in efforts to stem the spread and to reduce the effects of the disease. Our study can serve as an example of how, with systematic sampling, we can set conservation guidelines with the aim of preventing the further introductions and stemming the spread of Bd within a system.

204 Table Table 6.1: Details of populations of Sardinian amphibians sampled in the 2007 field-season. E. p. = Euproctus platycephalus , H. s. = Hyla sarda , D. s. = Discoglossus sardus , S. i. = Speleomantes imperialis, R/P = Rana/Pelophylax complex

% Elevation Region Site Coordinates n positive GE+/- S.E. (m) Sette Fratelli M1) Rio Marani N39° 20' 51.2" E9° 26' 27.6" 30 ( E. p. ) 0 n/a 162 Sette Fratelli M2) Forestry station N39° 16' 26.7" E9° 24' 58.4" 26 (10 H. s ., 16 D. s .) 0 n/a 517 Sette Fratelli M3) Rio Sa Ceraxa N39° 16' 22.0" E9° 26' 29.1" 25 ( E. p. ) 0 n/a 579 Sette Fratelli M4) Sopra Burcei N39° 21' 52.5" E9° 20' 39.0" 37 ( E. p. ) 0 n/a 262 Sette Fratelli M5) Speleomantes cave N39° 17' 42.8" E9° 26' 46.2" 61 ( S. i. ) 0 n/a 769 Sette Fratelli M6) Su Baccu de is Angiulus N39° 20' 21.9" E9° 28' 01.3" 30 ( E. p. ) 0 n/a 136 Sette Fratelli M7) Maidopis system N39° 18' 08.9" E9° 24' 15.4" 19 ( E. p.) 0 n/a 410 Limbara A1) Li Reni N40° 52' 30.6" E9° 10' 40.7" 14 (8 D. s. , 4 E. p. , 2 H. s. ) 0 n/a 749 Limbara A2) Fontana Fratelli N40° 52' 57.2" E9° 9' 48.8" 45 (44 H. s. , 1 D. s. ) 0 n/a 673 Limbara A3) Samela N40°49' 20.1" E9° 9' 54.9" 48 (43 H. s ., 5 E. p .) 0 n/a 676 Limbara A4) Su Bullone N40°46' 47.3" E8° 52' 30.2" 32 ( R/P ) 0 n/a 153 Limbara A5) Pisciaroni N40° 51' 42.1" E9° 08' 16.4" 30 ( H. s .) 0 n/a 716 Limbara A6) Legata Fontana N40° 48' 59.5" E9° 8' 26.7" 32 (30 H. s. , 2 D. s. ) 0 n/a 643 Limbara A7) Laghetto dei Pompieri N40° 50' 20.0" E9° 08' 35.5" 22 (17 D. s ., 5 H. s. ) 13.6 31.36 +/-16.56 983 Limbara A8) Culumbanu N40° 50' 30.3" E9° 06' 50.9" 53 ( E. p. ) 0 n/a 983 Gennargentu N1) Pischina Urtadala N40° 10' 22.3" E9° 29' 17.25" 3 ( E. p. ) 0 n/a 720

Gennargentu N2) Iseriagos saminna N39° 59' 22.4 E9° 18' 34.3" 2 ( E. p. ) 0 n/a 1279

Gennargentu N3) Perdas de Fogu N39° 36' 12.6" E9° 29' 29.6" 28 ( E. p. ) 0 n/a 486

Limbara Monte Olia N40° 44’ 45”, E9° 21’ 40” 1D. s. 100 66.70+/-14.6 -

Gennargentu Su Gunnue e Sebba N40° 10’ 20.28” E9° 29’ 20529.25” 87 (E. p.) 0 n/a 625 Figures

Figure 6.1: Locations of all populations sampled in the 2007 field season.

Blue denotes the Sette Fratelli region, red denotes the Gennargentu region, and yellow denotes the Limbara region.

206

Figure 6.2: Locations of sampled populations in the southern Sette Fratelli mountain region.

207 Figure 6.3: One of 10 specimens of Euproctus platycephalus with damaged digits collected at site M1, Rio Marani.

208 Figure 6.4: Locations of sampled populations in the northern Limbara mountain region.

209 Figure 6.5: Locations of positive sites in the Limbara mountain region.

210 Figure 6.6: Locations of sampled populations in the Gennargentu mountain region.

211 Chapter 7: General Conclusions

Given the finite resources available to tackle the current biodiversity crisis, it is important that the available time and money are allocated in a cost-effective manner. Many approaches have been advocated for doing just that (Brooks et al. 2006). At a species-level, prioritisation schemes may tally the number of species to obtain a spatial measure of species richness (e.g. Myers et al.

2000) or endemism (BirdLife International 2004a). The information obtained is then used to highlight ‘hotspots’ where we may focus conservation attention in order to protect as many species as possible, and maximise the ‘bang for our buck’. Although raw counts of species are informative, other information, such as the level of a species’ extinction risk (Alliance for Zero Extinction

2005) or evolutionary history (Isaac et al. 2007; Redding and Mooers 2006), may also be taken into account in order to highlight particularly vulnerable or irreplaceable species or sites. However, current schemes focussing on species vulnerability generally use spatial and environmental variables as a surrogate of habitat degradation in order to quantify the intensity of threat

(Brooks et al. 2006). There are a number of drawbacks to the use of habitat loss as a measure of vulnerability of a region: it is retrospective rather than proactive in nature; it does not take account of the variety of threat processes that may be acting upon species within a system; and it does not take into account the varying susceptibility of different species. An alternative approach, which addresses these problems, is to use information on a species’ biology, environment, and the intensity of threat experienced, to predict susceptibility to decline and an increase in threat status in the future,

212 and highlight those species as being suitable for pre-emptive conservation action (e.g. Cardillo et al. 2006).

I have focussed my doctoral research on gaining a better understanding of the observed patterns of amphibian extinction risk, with the aim of using this knowledge to predict species’ susceptibility to chytridiomycosis and direct subsequent conservation efforts. In doing so, my work has contributed new insights into how comparative studies may be used to inform the direction of cost-effective, focussed research and applied conservation.

In the first empirical chapter of my thesis (chapter 2), I assessed how heterogeneity in threat intensity, knowledge of conservation status, and species biology underpins the taxonomic distribution of extinction risk in amphibians. Although these mechanisms have been discussed at length in the literature (Russell et al. 1998), to my knowledge this analyses represents the first explicit test of their role in selectivity in extinction risk. My analyses suggest that all three mechanisms contribute to the observed pattern. The implications of this are two-fold. Firstly, the treatment of data deficient species should be carefully considered before analyses are conducted or status reports are prepared (e.g. Baillie et al. 2004; Stuart et al. 2004). A failure to do so may result in inaccurate conclusions regarding the proportion of threatened species recorded at higher taxonomic levels, and those conclusions may not be restricted to poorly studied taxa with a high proportion of DD species.

Second, the role of biology in determining a species level of risk indicates that conservation actions based on threat intensity alone is unlikely to be effective;

213 the biology of species should be explicitly considered in conservation decision-making.

If biology is important in rendering a species more or less susceptible to threatening processes, it is important to understand which biological traits increase the susceptibility of a species. Although several studies have looked at associations between traits and declines at a local or regional level in amphibians (Lips et al. 2003; Murray and Hose 2005; Williams and Hero

1998), my third chapter represents the first published study of such associations in amphibians that have experienced an increase in extinction risk at the global level using the phylogenetic comparative method. Although more focussed, regional analyses may be particularly insightful for directing conservation management of a specific system (Fisher and Owens 2004), when regional or even population level idiosyncrasies in response to a threat exist, global analyses can highlight overall trends and commonalities that are more likely to be transferable among regions. Chapter 3 also takes a further step by using the output of the most predictive model to highlight which species are most likely to be susceptible to future declines as a result of infection by Batrachochytrium dendrobatidis ( Bd ). Although such predictions are based on extrapolations from past trends, and therefore assume that current conditions and trends remain linear and constant, they can be very effective in highlighting robust trends in the data with a view to making predictions on what will happen next, and directing conservation resources accordingly (Sutherland 2006). While some of the locations I identified with a large number of susceptible species are already the subject of Bd sampling

214 (e.g. the Andes), others, such as the Western Ghats in India, New Guinea, and southern China have received little if any sampling as yet. These locations therefore represent a high priority for future efforts to sample for Bd and to monitor populations (Collins and Halliday 2005).

The model presented in chapter 3 represents a starting point in predicting species susceptibility to Bd infection. It is important that if models are to be useful in directing conservation, they are fine-tuned and improved as our knowledge of both the pathogen and hosts improves. The incorporation of more amphibian ecology and life-history traits would be useful in gaining a better understanding of the role of species biology in susceptibility to decline.

Unfortunately. at the global level, the lack of such biological information has limited the number of traits and species included in the analyses. In the short- term, while more information on amphibian biology in certain regions is lacking, more focussed models on regional systems, with finer scales of decline may be useful for optimising the predictor and response variables that we use and in improving the accuracy of predictions of species’ susceptibility.

As well as improving the quality and quantity of the information included in the model, it is also important to consider factors that may affect a species response but that are presently absent from the model (see Mustin et al.

2007). In the case of the model presented in chapter three, species susceptibility is likely to be affected by additional factors, such as a species immune defences (Woodhams et al. 2007); variation in the virulence of the Bd strain that is present (Berger et al. 2005b), and Bd ’s biological or

215 environmental requirements (Ron 2005). Efforts to include these extra sources of variation would be an interesting extension of the work presented here. Indeed, I am currently involved in a project to combine the predictions made on Bd susceptibility with niche models that predict the pathogen’s potential distribution.

If predictive models of extinction risk are to be useful for practical conservation, we need to ensure that we use the best methods available.

Chapter four of my thesis assessed the use of three techniques that have been advocated for modelling extinction risk, providing the first systematic comparison of regression techniques, both with and without incorporating phylogeny, and decision trees. On the basis of my results, future attempts to model extinction risk should incorporate a suitable phylogenetic comparative method to account for phylogenetic non-independence and to increase the precision of the predictions made. However, decision trees may prove useful in preliminary analyses, given their potential to highlight complex relationships among variables, which can then be specified as interaction terms in the phylogenetic comparative analyses.

An ultimate aim of predictive modelling in ecology and conservation is to make predictions that others, whether they be researchers or conservation practitioners, can use (Sutherland 2006). My fifth and sixth chapters illustrate a simple way in which we can use these predictions to direct further research, specifically sampling for Bd . So far, much of the Bd sampling conducted for research purposes has been based around long-term study sites or species

216 (e.g. Bosch et al. 2001; e.g. Lips et al. 2006; Rachowicz et al. 2006) or focussed on a particular geographic region (e.g. Garner et al. 2005). In chapter five I highlight how, using the output of the model of Bd susceptibility from chapter three, model predictions can direct sampling of target species with the goal of expanding global Bd surveillance in a more systematic and cost-effective manner. This is of particular importance in regions lacking any current sampling efforts such as South and South-East Asia, which are too large for systematic geographic surveillance (e.g. chapter six). The procedure presented here highlighted one species that, given its declining population and the presence of Bd, certainly warrants further, more focussed, research with regards its population status and the role of chytridiomycosis in its decline. Ideally, the targeted surveillance presented here would have been even more focussed in terms of the number of samples taken to detect Bd presence. Several procedures exist in the literature for calculating the minimum sample size required to gain an accurate impression of the presence of a trend or pattern (e.g. Baillie et al. In Press; Franco et al. 2007). If such a procedure could be developed for Bd sampling and compared and contrasted with current epidemiological models (e.g. Thrusfield 2005) we may be able to increase the cost-effectiveness of Bd sampling in new regions.

Chapter six described one such piece of focussed research, which assessed the distribution of Bd on the island of Sardinia, one of the most endemic and highly threatened amphibian assemblages in Europe. This chapter stands as an example of how, for a relatively small amount of time and money, directed research can be used to recommend a number of future directions for the

217 management of Bd . The results of chapter six suggest that Bd is restricted in its distribution on the island and that, in addition to more research on the subject, efforts should be made to prevent further dissemination of the pathogen. In order to do so, collaboration between the local conservation practitioners and policy-makers is vital. Given the enthusiasm and interest that the forestry workers and managers of Sardinia showed in this research, it seems likely that further plans to involve the forestry department in this work in order to set conservation targets would be welcomed, and this is something that my collaborators on the island and I are actively pursuing. Given the importance of increasing the flow of information between research and conservation practitioners and policy-makers, such collaboration can only be a positive move for the conservation of Sardinia’s amphibian species.

Amphibian conservation faces some major challenges in the near future. The conversion of academic research into applied conservation is one such challenge, and this is an area I aimed to tackle in my doctoral research. In doing so, I have illustrated one method of prioritising species for conservation action, and highlight how the model output may be used to direct conservation management and policy. However, amphibian conservation faces some major challenges in how it deals with the threat caused of Bd in the future. Improving our knowledge of the basic biology and population trends of susceptible species is an obvious, yet important, area in which we need to invest time and money if we are to accurately monitor the future loss of amphibian populations, and ascertain which species are most susceptible to those declines (Collins and Halliday 2005). In conjunction with better

218 knowledge of biology and population status, Bd surveillance in poorly-known regions would help efforts to understand species response to Bd exposure, infection, and disease occurrence.

For populations suffering declines as a result of Bd infection, the ultimate objective must be to find an effective treatment for the pathogen at a landscape level. At present, although individually infected animals can be treated in captivity, no suitable wide-scale applications exist. In the interim, the aim of many conservation organisations and animal collections is to establish ex situ collections of animals with the aim of reintroduction at a later date (http://www.amphibianark.org/ 2008). However, landscape level treatment and ex situ management, are major challenges that are unlikely to have a silver-bullet solution; many species and sites may have idiosyncrasies and special requirements. It is therefore important that at each step along the way there is clear communication between field-researchers, and conservation practitioners and managers in order to improve the flow of information on species population trends, infection status, response to infection, management of ex situ populations, and landscape level trials.

Several large-scale, collaborative projects are already in place (e.g. http://www.amphibianark.org/ 2008; http://www.spatialepidemiology.net/bd), which may be extended in size and scope to share knowledge that may otherwise not reach the published literature (e.g. species response to exposure to Bd , or species optimal ex situ husbandry). Additionally the use of adaptive management regimes, and the communal databases of conservation successes and failures (Sutherland et al. 2004) may play an important role in

219 combating future declines and preventing further extinctions as a result of chytridiomycosis.

220 References

ACAP. 2005. Amphibian Conservation Summit, 17–19 September 2005,

Washington DC.

Akçakaya, H. R., S. Ferson, M. A. Burgman, D. A. Keith, G. M. Mace, and C.

R. Todd. 2000. Making consistent IUCN classifications under

uncertainty. Conservation Biology 14:1001-1013.

Alexander, M. A., and J. K. Eischeid. 2001. Climate Variability in regions of

Amphibian Declines. Conservation Biology 15:930-942.

Alliance for Zero Extinction. 2005. Criteria for definition of AZE sites.

Amiet, J.-L. 1975. Ecologie et distribution des Amphibiens Anoures de la

region de Nkongsamba (Cameroun). Annales Faculte des Sciences de

Yaounde 20:33-107.

Andreone, F., and L. Luiselli. 2000. The Italian batrachofauna and its

conservation status:a statistical assessment. Biological Conservation

96:197-208.

Araujo, M. B., R. G. Pearson, W. Thuiller, and M. Erhard. 2005a. Validation of

species-climate impact models under climate change. Global Change

Biology 11:1504-1513.

Araujo, M. B., W. Thuiller, and R. G. Pearson. 2006. Climate warming and the

decline of amphibians and reptiles in Europe. Journal of Biogeography

33:1712-1728.

Araujo, M. B., R. J. Whittaker, R. J. Ladle, and M. Erhard. 2005b. Reducing

uncertainty in projections of extinction risk from climate change. Global

Ecology and Biogeography 14:529-538.

221 Australian Government. 2006. Threat abatement plan: Infection of amphibians

with chytrid fungus resulting in chytridiomycosis., Pages 21. Canberra,

Australian Government.

Baillie, J. E. M., B. Collen, R. Amin, H. R. Akcakaya, S. Butchart, N. Brummitt,

T. R. Meagher et al. In Press. Toward monitoring global biodiversity.

Conservation Letters.

Baillie, J. E. M., C. Hilton-Taylor, and S. N. Stuart. 2004. 2004 IUCN Red List

of Threatened Species. A Global Species Assessment., IUCN, Gland,

Switzerland and Cambridge, UK.

Balmford, A. 1996. Extinction filters and current resilience: The significance of

past selection pressures for conservation biology. Trends in Ecology &

Evolution 11:193-196.

Beard, K. H., and E. M. O'Neill. 2005. Infection of an invasive frog

Eleutherodactylus coqui by the chytrid fungus Batrachochytrium

dendrobatidis in Hawaii. Biological Conservation 126:591-595.

Bennett, P. M., and I. P. F. Owens. 1997. Variation in extinction risk among

birds: Chance or evolutionary predisposition? Proceedings of the Royal

Society of London Series B-Biological Sciences 264:401-408.

Berger, L., A. Hyatt, R. Speare, and J. E. Longcore. 2005a. Life cycle stages

of the amphibian chytrid, Batrachochytrium dendrobatidis . Diseases of

Aquatic Organisms 68:51-63.

Berger, L., G. Marantelli, L. F. Skerratt, and R. Speare. 2005b. Virulence of

the amphibian chytrid fungus Batrachochytrium dendrobatidis varies

with the strain. . Diseases of Aquatic Organisms 68:47-50.

222 Berger, L., R. Speare, P. Daszak, D. E. Green, A. A. Cunningham, C. L.

Goggin, R. Slocombe et al. 1998. Chytridiomycosis causes amphibian

mortality associated with population declines in the rain forests of

Australia and Central America. Proceedings of the National Academy

of Sciences of the United States of America 95:9031-9036.

Berger, L., R. Speare, H. Hines, G. Marantelli, A. Hyatt, K. R. McDonald, L. F.

Skerratt et al. 2004. Effects of season and temperature on mortality in

amphibians due to chytridiomycosis. Australian Veterinary Journal

82:31-36.

Bielby, J., N. Cooper, A. A. Cunningham, T. W. J. Garner, and A. Purvis.

2008. Predicting rapid declines in the world's frogs. Conservation

Letters In press.

Bielby, J., A. A. Cunningham, and A. Purvis. 2006. Taxonomic selectivity in

amphibians: ignorance, geography or biology? Animal Conservation

9:135-143.

Bininda-Emonds, O. R. P., M. Cardillo, K. E. Jones, R. D. E. MacPhee, R. M.

D. Beck, R. Grenyer, S. A. Price et al. 2007. The delayed rise of

present-day mammals. Nature 446:507-512.

BirdLife International. 2004a. BirdLife International EBA prorgramme.

—. 2004b. Important Bird Areas.

—. 2007. BirdLife's online World Bird Database: the site for bird conservation.

in V. 2.1., ed. Cambridge, UK: BirdLife International.

Blaustein, A. R., and D. B. Wake. 1990. Declining Amphibian Populations - a

Global Phenomenon. Trends in Ecology & Evolution 5:203-204.

223 Bonnett, X., R. Shine, and O. Lourdais. 2002. Taxonomic chauvinism. Trends

in Ecology & Evolution 17:1-3.

Bosch, J., L. M. Carrascal, L. Duran, S. Walker, and M. C. Fisher. 2007.

Climate change and outbreaks of amphibian chytridiomycosis in a

montane area of Central Spain; is there a link? Proceedings of the

Royal Society B-Biological Sciences 274:253-260.

Bosch, J., and I. Martinez-Solano. 2006. Chytrid fungus infection related to

unusual mortalities of Salamandra salamandra and Bufo bufo in the

Penalara Natural Park, Spain. Oryx 40:84-89.

Bosch, J., I. Martinez-Solano, and M. Garcia-Paris. 2001. Evidence of a

chytrid fungus infection involved in the decline of the common midwife

toad (Alytes obstetricans) in protected areas in central Spain. Biological

Conservation 97:331-337.

Bovero, S., G. Sotgiu, C. Angelini, S. Doglio, E. Gazzaniga, and L. Picciau.

2005. New data on the distribution of the Sardinian brook salamander

(Euproctus platycephalus ) in the southern and western Limbara

mountain complex (Sardinia). Herpetological Bulletin 93.

Bovero, S., G. Sotgui, C. Angelini, S. Doglio, E. Gazzaniga, A. A.

Cunningham, and T. W. J. Garner. in press. Detection of

chytridiomycosis caused by Batrachochytrium dendrobatidis in the

endangered Sardinian Newt Euproctus platycephalus in southern

Sardinia, Italy. Journal of Wildlife Disease.

Boyle, D. G., D. B. Boyle, V. Olsen, J. A. T. Morgan, and A. D. Hyatt. 2004.

Rapid quantitative detection of chytridiomycosis ( Batrachochytrium

224 dendrobatidis ) in amphibian samples using real-time Taqman PCR

assay. Diseases of Aquatic Organisms 60:141-148.

Breiman, L., J. H. Friedman, R. A. Olshen, and C. G. Stone. 1984,

Classification and Regression Trees. Belmont, California, Wadsworth

International Group.

Brooks, T., G. A. B. da Fonseca, and A. S. L. Rodrigues. 2004. Protected

areas and species. Conservation Biology 18:616-618.

Brooks, T. M., R. A. Mittermeier, G. A. B. da Fonseca, J. Gerlach, M.

Hoffmann, J. F. Lamoreux, C. G. Mittermeier et al. 2006. Global

Biodiversity Conservation Priorities. Science 313:58-61.

Brown, J. H. 1995, Macroecology . Chicago, Chicago University Press. Burrowes, P. A., R. L. Joglar, and D. E. Green. 2004. Potential causes for

amphibian declines in Puerto Rico. Herpetologica 60:141-154.

Butchart, S. H. M., H. R. Akçakaya, J. Chanson, J. E. M. Baillie, B. Collen, S.

Quader, W. R. Turner et al. 2007. Improvements to the Red List Index.

. PLoS One 2:e140.

Butchart, S. H. M., H. R. Akcakaya, E. Kennedy, and C. Hilton-Taylor. 2006.

Biodiversity Indicators Based on Trends in Conservation Status:

Strengths of the IUCN Red List Index. Conservation Biology 20:579-

581.

Butchart, S. H. M., A. J. Stattersfield, L. A. Bennun, S. M. Shutes, H. R.

Akcakaya, J. E. M. Baillie, S. N. Stuart et al. 2004. Measuring global

trends in the status of biodiversity: Red list indices for birds. Plos

Biology 2:2294-2304.

Cardillo, M., G. M. Mace, J. L. Gittleman, K. E. Jones, J. Bielby, and A. Purvis.

2008. The predictability of extinction: biological and external correlates

225 of decline in mammals. Proceedings of the Royal Society B-Biological

Sciences In review.

Cardillo, M., G. M. Mace, J. L. Gittleman, and A. Purvis. 2006. Latent

extinction risk and the future battleground of mammal conservation.

Proceedings of the National Academy of Sciences of the United States

of America 103:4157-4161.

Cardillo, M., G. M. Mace, K. E. Jones, J. Bielby, O. R. P. Bininda-Emonds, W.

Sechrest, C. D. L. Orme et al. 2005a. Multiple causes of high extinction

risk in large mammal species. Science 309:1239-1241.

Cardillo, M., G. M. M. Mace, and A. Purvis. 2005b. Response: Problems of

studying extinction risks. Science 310:1277-1278.

Cardillo, M., A. Purvis, W. Sechrest, J. L. Gittleman, J. Bielby, and G. M.

Mace. 2004. Human population density and extinction risk in the

world's carnivores. Plos Biology 2:909-914.

Carruccio, A. 1869. Catalogo metodico degli animali vertebrati riportati dalle

escursioni nelle province meridionali, in Sicilia e in Sardegna, negli

anni1868–1869. . Atti Soc. Ital. Sci. nat. 12:553-586.

CIESIN. 2000. Gridded Population of the World

(GPW), Version

2. Palisades. NY, CIESIN, Columbia University.

Clough, M., and K. Summers. 2000. Phylogenetic systematics and

biogeography of the poison frogs: evidence from mitochondrial DNA

sequences. Biological Journal of the Linnean Society 70:515-540.

226 Collen, B., E. Bykova, S. Ling, E. J. Milner-Gulland, and A. Purvis. 2006.

Extinction risk: A comparative analysis of central Asian vertebrates.

Biodiversity and Conservation 15:1859-1871.

Collins, J. P., and T. Halliday. 2005. Forecasting changes in amphibian

biodiversity: aiming at a moving target. Philosophical Transactions of

the Royal Society B-Biological Sciences 360:309-314.

Collins, J. P., and A. Storfer. 2003. Global amphibian declines: sorting the

hypotheses. Diversity and Distributions 9:89-98.

Colomo, S. 1999. Anfibi. in A. F. Sardo, ed. In: Guida pratica alla Flora e alla

Fauna della Sardegna. Nuoro.

Conservation International. 2005. Biodiversity Hotspots: Species Database.

Cooper, N., J. Bielby, G. Thomas, and A. Purvis. 2008. Macroecology and

exinction risk correlates of frogs. Global Ecology and Biogeography

17:211-221.

Crawley, M. J. 2002, Statistical Computing - An Introduction to Data Analysis

using S-Plus, Wiley.

Cunningham, A. A., A. Hyatt, P. Russell, and P. M. Bennett. 2007. Emerging

epidemic diseases of frogs in Britain are dependent on the source of

ranavirus agent and the route of exposure. Epidemiology and Infection

135:1200-1212.

Cunningham, A. A., T. E. S. Langton, P. M. Bennett, J. F. Lewin, S. E. N.

Drury, R. E. Gough, and S. K. MacGregor. 1996. Pathological and

microbial findings from incidents of unusual mortality of the common

frog ( Rana temporaria ). Philosophical Transactions of the Royal

Society B-Biological Sciences 351:1539-1557.

227 Daly, G. L., Y. D. Lei, C. Teixeira, D. C. G. Muir, L. E. Castillo, and F. Wania.

2007. Accumulation of current-use pesticides in Neotropical montane

forests. Environmental Science and Technology 41:1118-1123.

Damuth, J. 1987. Interspecific allometry of population density in mammals and other animals: the independence of body mass and population energy- use. Biological Journal of the Linnean Society 31:193-246. Daszak, P., L. Berger, A. A. Cunningham, A. D. Hyatt, D. E. Green, and R.

Speare. 1999. Emerging Infectious Diseases and Amphibian

Population Declines. Emerging Infectious Diseases 5:735-748.

Daszak, P., A. A. Cunningham, and A. D. Hyatt. 2003. Infectious disease and

amphibian population declines. Diversity and Distributions 9:141-150.

Daszak, P., D. E. Scott, A. M. Kilpatrick, C. Faggioni, J. W. Gibbons, and D.

Porter. 2005. Amphibian population declines at savannah river site are

linked to climate, not chytridiomycosis. Ecology 86:3232-3237.

Daszak, P., A. Strieby, A. A. Cunningham, J. E. Longcore, C. C. Brown, and

D. Porter. 2004. Experimental evidence that the bullfrog ( Rana

catesbeiana ) is a potential carrier of chytridiomycosis, an emerging

fungal disease of amphbians. Herpetological Journal 14:201-207.

Davidson, C. 2004. Declining downwind: Amphibian population declines in

california and historical pesticide use. Ecological Applications 14:1892-

1902.

Davidson, C., M. F. Bernard, H. B. Shaffer, J. M. Parker, C. O'Leary, J. M.

Conlon, and L. A. Rollins-Smith. 2007. Effects on Chytrid and Carbaryl

Exposure on Survival, Growth and Skin Peptide Defenses in Foothill

Yellow-legged Frogs. Environmental Science Technology 41:1771-

1776.

228 De'ath, G., and K. E. Fabricius. 2000. Classification and regression trees: a

powerful yet simple technique for ecological data analysis. Ecology

81:3178-3192. de Castro, F., and B. Bolker. 2005. Mechanism of disease-induced extinction.

Ecology Letters 8:117-126.

De Grammont, P. C., and A. D. Cuaron. 2006. An Evaluation of Threatened

Specis Categorization Systems Used on the American Continent.

Conservation Biology 20:14-27.

Dodds, C. K., and R. A. Seigel. 1991. Relocation, repatriation, and

translocation of amphibians and reptiles: are they conservation

strategies that work? Herpetologica 47:336-350.

Donnelly, M. A. 1994. Amphibians of La Selva, Pages 380-381 in L. A.

McDade, K. S. Bawa, H. A. Hespenheide, and G. S. Hartshorn, eds. La

Selva: ecology and natural history of a neotropical rain forest. Chicago,

University of Chicago Press.

Elith, J., C. H. Graham, R. P. Anderson, M. Dudik, S. Ferrier, A. Guisan, R. J.

Hijmans et al. 2006. Novel methods improve prediction of species'

distributions from occurrence data. Ecography 29:129-151.

Felsenstein, J. 1985. Phylogenies and the Comparative Method. American

Naturalist 125:1-15.

Fisher, D. O., and I. P. F. Owens. 2004. The comparative method in

conservation biology. Trends in Ecology & Evolution 19:391-398.

Fisher, M. C., and T. W. J. Garner. 2007. The relationship between the

introduction of Batrachochytrium dendrobatidis , the international trade

229 in amphibians and introduced amphibian species. Fungal Biology

Reviews 21:2-9.

Fox, S. F., A. L. Greer, R. Torres-Cervantes, and J. P. Collins. 2006. First

case of ranavirus-associated morbidity and mortality in natural

populations of the South American frog Atelognathus patagonicus .

Diseases of Aquatic Organisms 72:87-92.

Franco, A. M. A., J. M. Palmeirim, and W. J. Sutherland. 2007. A method for

comparing effectiveness of research techniques in conservation and

applied ecology. Biological Conservation 134:96-105.

Freckleton, R. P., P. H. Harvey, and M. Pagel. 2002. Phylogenetic analysis

and comparative data: A test and review of evidence. American

Naturalist 160:712-726.

Frost, D. R. 2004. Amphibian Species of the World: an Online Reference.

Version 3.0 [22 August, 2004]).

—. 2007. Amphibian Species of the World: an Online Reference. Version 5.0

(1 February, 2007). Electronic Database accessible at

http://research.amnh.org/herpetology/amphibia/index.php . in A. M. o.

N. History, ed. New York, USA.

Frost, D. R., T. Grant, J. Faivovich, R. H. Bain, A. Hass, C. F. B. Haddad, R.

O. De Sa et al. 2006. The Amphibian Tree of Life. Bulletin of the

American Museum of Natural History 297:1-370.

Garland, T., P. H. Harvey, and A. R. Ives. 1992. Procedures for the analysis of

comparative data using phylogenetically independent contrasts.

Systematic Biology 41:18-32.

230 Garner, T. W. J., M. Perkins, P. Govindarajulu, D. Seglie, S. F. Walker, A. A.

Cunningham, and M. C. Fisher. 2006. The emerging amphibian

pathogen Batrachochytrium dendrobatidis globally infects introduced

populations of the North American bullfrog, Rana catesbeiana. Biology

Letters 2:455-459.

Garner, T. W. J., S. Walker, J. Bosch, A. D. Hyatt, A. A. Cunningham, and M.

C. Fisher. 2005. Chytrid fungus in Europe. Emerging Infectious

Diseases 11:1639-1641.

Gene, J. 1839. Synopis Reptilium Sardiniae indigenorum. Mem. Accad. Sci.

Torino. Ser. 2:257-286.

Gittleman, J. L., and H. K. Luh. 1992. On comparing comparative methods.

Annual Review of Ecology and Systematics 23:383-404.

Goldberg, T. L., A. M. Readel, and M. H. Lee. 2007. Chytrid fungus in frogs

from an equatorial African montane forest in West Uganda. Journal of

Wildlife Disease 43:521-524.

Grafen, A. 1989. The phylogenetic regression. Philosophical Transactions of

the Royal Society of London B, Biological Sciences 326:119-157.

Green, D. E., K. A. Converse, and A. K. Schrader. 2002. Epizootiology of

sixty-four amphibian morbidity and mortality events in the USA, 1996-

2001. Annals of the New York Academy of Sciences. 969:323-339.

Grenyer, R., C. D. L. Orme, S. F. Jackson, G. H. Thomas, R. G. Davies, T. J.

Davies, K. E. Jones et al. 2006. Global distribution and conservation of

rare and threatened vertebrates. Nature 444:93-96.

231 Halsey, L. G., T. M. Blackburn, and P. J. Butler. 2006. A comparative analysis

of the diving behaviour of birds and mammals. Functional Ecology

20:889-899.

Harmon, L. J., and J. B. Losos. 2005. The effect of intraspecific sample size

on Type I and Type II error rates in comparative studies. Evolution

59:2705-2710.

Harvey, P. H., and M. Pagel. 1991, The comparative method in evolutionary

biology. Oxford, Oxford University Press.

Harvey, P. H., and A. Rambaut. 1998. Phylogenetic extinction rates and

comparative methodology. Proceedings of the Royal Society of London

Series B-Biological Sciences 265:1691-1696.

Hero, J. M., S. E. Williams, and W. E. Magnusson. 2005. Ecological traits of

declining amphibians in upland areas of eastern Australia. Journal of

Zoology 267:221-232.

Herrmann, H.-W., W. Bohme, P. A. Herrmann, M. Plath, A. Schmitz, and M.

Solbach. 2005. African biodiversity hotspots: the amphibians of Mt.

Nlonako, Cameroon. Salamandra 41:61-81.

Hijmans, R. J., S. E. Cameron, J. L. Parra, P. G. Jones, and A. Jarvis. 2005.

Very high resolution interpolated climate surfaces for global land

areas.. International Journal of Climatology 25:1965-1978.

Houlahan, J. E., C. S. Findlay, B. R. Schmidt, A. H. Meyer, and S. L. Kuzmin.

2000. Quantitative evidence for global amphibian population declines.

Nature 404:752-755. http://www.amphibianark.org/ . 2008. http://www.spatialepidemiology.net/bd .

232 Hyatt, A. D., D. G. Boyle, V. Olsen, D. B. Boyle, L. Berger, D. Obendorf, A.

Dalton et al. 2007. Diagnostic assays and sampling protocols for the

detection of Batrachochytrium dendrobatidis . Diseases of Aquatic

Organisms 73:175-192.

Ihaka, R., and R. Gentleman. 1996. R:A language for data analysis and

graphics. Journal of Computational and Graphical Statistics 5:299-314.

Isaac, N. J. B., J. Mallet, and G. M. Mace. 2004. Taxonomic inflation: its

influence on macroecology and conservation. Trends in Ecology &

Evolution 19:464-469.

Isaac, N. J. B., S. T. Turvey, B. Collen, C. Waterman, and J. Baillie. 2007.

Mammals on the EDGE: Conservation Priorities Based on Threat and

Phylogeny. PLos One 3.

IUCN. 2001. IUCN Red List Categories and Criteria: Version 3.1. Gland,

Switzerland and Cambridge, UK, IUCN.

—. 2004. 2004 IUCN Red List of Threatened Species.

—. 2008. IUCN Red List.

IUCN, Conservation International, and NatureServe. 2004. Global Amphibian

Assessment. < www.globalamphibians.org > Downloaded January

2008.

Ives, A. R., P. E. Midford, and T. Garland. 2007. Within-species variation and

measurement error in phylogenetic comparative methods. Systematic

Biology 56:252-270.

Johnson, B. 1999. Recovery of the Puerto Rican crested toad. Endangered

Species Update 24:8-9.

233 Johnson, M. L., L. Berger, L. Philips, and R. Speare. 2003. Fungicidal effects

of chemical disinfectants, UV light, desiccation and heat on the

amphibian chytrid Batrachochytrium dendrobatidis . Diseases of Aquatic

Organisms 57:255-260.

Johnson, M. L., and R. Speare. 2003. Survival of Batrachochytrium

dendrobatidis in Water: Quarantine and Disease Control Implications.

Emerging Infectious Diseases 9:922-925.

—. 2005. Possible modes of dissemination of the amphibian chytrid

Batrachochytrium dendrobatidis in the environment. Diseases of

Aquatic Organisms 65:181-186.

Jones, K. E., A. Purvis, and J. L. Gittleman. 2003. Biological correlates of

extinction risk in bats. American Naturalist 161:601-614.

Jones, M. J., A. Fielding, and M. Sullivan. 2006. Analysing extinction risk in

parrots using decision trees. Biological Conservation 15:1993-2007.

Kats, L. B., and R. P. Ferrer. 2003. Alien predators and amphibian declines:

review of two decades of science and the transition to conservation.

Diversity and Distributions 9:99-110.

Kerr, J. T., and D. J. Currie. 1995. Effects of human activity on global

extinction risk. Conservation Biology 9:1528-1538.

Kiesecker, J. M., A. R. Blaustein, and L. K. Belden. 2001. Complex causes of

amphibian population declines. Nature 410:681-684.

Koh, L. P., N. S. Sodhi, and B. W. Brook. 2004. Ecological Proneness of

Extinction Proneness in Tropical Butterflies. Conservation Biology

18:1571-1578.

234 Kotiaho, J. S., V. Kaitala, A. Komonen, and J. Paivinen. 2005. Predicting the

risk of extinction from shared ecological characteristics. Proceedings of

the National Academy of Sciences 102:1963-1967.

Kriger, K. M., and J. M. Hero. 2007. Large-scale seasonal variation in the

prevalence and severity of chytridiomycosis. Journal of Zoology

271:352-359.

Kriger, K. M., J. M. Hero, and K. J. Ashton. 2006a. Cost efficiency in the

detection of chytridiomycosis using PCR assay. Diseases of Aquatic

Organisms 71:149-154.

Kriger, K. M., H. B. Hines, A. D. Hyatt, D. G. Boyle, and J. M. Hero. 2006b.

Techniques for detecting chytridiomycosis in wild frogs: comparing

histology with real-time Taqman PCR. Diseases of Aquatic Organisms

71:141-148.

La Marca, E., K. R. Lips, S. Lötters, R. Puschendorf, R. Ibanez, J. V. Rueda-

Almonacid, R. Schulte et al. 2005. Catastrophic population declines

and extinctions in neotropical harlequin frogs (Bufonidae: Atelopus ).

Biotropica 37:190-201.

Laurance, W. F. 1991. Ecological Correlates of Extinction Proneness in

Australian Tropical Rain Forest Mammals. Conservation Biology 5:79-

89.

Laurance, W. F., K. R. McDonald, and R. Speare. 1996. Epidemic disease

and the catastrophic decline of Australian rain forest frogs.

Conservation Biology 10.

Lawson, D. P. 1993. The reptiles and amphibians of Korup National Park

project, Cameroon. Herpetological Natural History 1:27-90.

235 Lecis, R., and K. Norris. 2003. Habitat correlates of distribution and local

population decline of the endemic Sardinian newt Euproctus

platycephalus . Biological Conservation 115:303-317.

—. 2004. Population genetic diversity of the endemic Sardinian newt

Euproctus platycephalus : implications for conservation. Biological

Conservation 119:263-270.

Lips, K. R. 1998. Decline of a tropical montane amphibian fauna.

Conservation Biology 12:106-117.

—. 1999. Mass mortality and population declines of anurans at an upland site

in western Panama. Conservation Biology 13:117-125.

Lips, K. R., F. Brem, R. Brenes, J. D. Reeve, R. A. Alford, J. Volyes, C. Carey

et al. 2006. Emerging infectious disease and loss of biodiversity in a

Neotropical amphibian community. Proceedings of the National

Academy of Sciences of the United States of America 103:3165-3170.

Lips, K. R., J. E. Diffendorfer, J. R. Mendelson, and M. W. Sears. 2008. Riding

the Wave: Reconciling the Roles of Disease and Climate Change in

Amphibians Declines. Plos Biology 6:e72.

doi:10.1371/journal.pbio.0060072.

Lips, K. R., J. D. Reeve, and L. R. Witters. 2003. Ecological Traits Predicting

Amphibian Population Declines in Central America. Conservation

Biology 17:1078-1088.

Longcore, J. E., A. P. Pessier, and J. D. Nichols. 1999. Batrachochytrium

dendrobatidis gen. et sp. nov., a chytrid pathogenic to amphibians.

Mycologia 91:219-227.

236 Mace, G. M. 2004. The role of taxonomy in species conservation.

Philosophical Transactions of the Royal Society B-Biological Sciences

359:711-719.

Mace, G. M., and A. Balmford. 2000. Patterns and processes in contemporary

mammalian extinction, Pages 27-52 in A. Entwistle, and N. Dunstone,

eds. Priorities for the Conservation of Mammalian Diversity.

Cambridge, Cambridge University Press.

Mace, G. M., H. Masundire, and J. E. M. Baillie. 2005. Biodiversity Millennium

Ecosystem Assessment, 2005 Current State and Trends: Findings of

the Condition and Trends Working Group Ecosystems and Human

Well-being, vol1,. . Washington DC, Island Press.

Malkmus, R., U. Manthey, G. Vogel, P. Hoffmann, and J. Kosuch. 2002,

Amphibians and Reptiles of Mount Kinabalu.

Marantelli, G., L. Berger, R. Speare, and L. Keegan. 2004. Distribution of the

amphibian chytrid Batrachochytrium dendrobatidis and keratin during

tadpole development. Pacific Conservation Biology 10:173-179.

MacArthur, R. H., and E. O. Wilson. 1967, The theory of island biogeography, Princeton University Press. McKinney, M. L. 1997. Extinction vulnerability and selectivity: combining

ecological and paleontological views. Annual Review of Ecology and

Systematics 28:495-516.

—. 2001. Role of human population in raising bird and mammal threat among

nations. Animal Conservation 4:45-57.

Mendelson, J. R., K. R. Lips, J. E. Diffendorfer, R. W. Gagliardo, G. B. Rabb,

J. P. Collins, P. Daszak et al. 2006a. Responding to amphibian loss -

Response. Science 314:1541-1542.

237 Mendelson, J. R., K. R. Lips, R. W. Gagliardo, G. B. Rabb, J. P. Collins, J. E.

Diffendorfer, P. Daszak et al. 2006b. Biodiversity - Confronting

amphibian declines and extinctions. Science 313:48-48.

Missouri Department of Conservation. 2007. Hellbender Recovery Actions in

Missouri, Science Note.

Morehouse, E. A., T. Y. James, A. R. D. Ganley, R. Vilgalys, L. Berger, P. J.

Murphy, and J. E. Longcore. 2003. Multilocus sequence typing

suggests the chytrid pathogen of amphibians is a recently emerged

clone. Molecular Ecology 12:395-403.

Morrison, C., and J. M. Hero. 2003. Geographic variation in life-history characteristics of amphibians: a review. Journal of Animal Ecology 72:270-279. Morrison, C., J. M. Hero, and J. Browning. 2004. Altitudinal variation in the age at maturity, longevity, and reproductive lifespan of anurans in subtropical queensland. Herpetologica 60:34-44. Moyer, D. C., and C. Weldon. 2006. Chytrid distribution and pathogenicity

among frogs of the Udzungwa Mountains, Tanzania.

Murray, B. R., and G. C. Hose. 2005. Life-history and ecological correlates of

decline and extinction in the endemic Australian frog fauna. Australian

Ecology 30:564-571.

Mustin, K., W. J. Sutherland, and J. A. Gill. 2007. The complexity of predicting

climate-induced ecological impacts. Climate Research 35:165-175.

Muths, E., P. S. Corn, A. P. Pessier, and D. E. Green. 2003. Evidence for a

disease related decline in Colorado. Biological Conservation 110:357-

365.

238 Myers, N., R. A. Mittermeier, C. G. Mittermeier, G. A. B. da Fonseca, and J.

Kent. 2000. Biodiversity hotspots for conservation priorities. Nature

403:853-858.

Ouellet, M., I. Mikaelian, B. D. Pauli, J. Rodrigue, and D. M. Green. 2005.

Historical Evidence of Widespread Chytrid Infection in North American

Amphibian Populations. Conservation Biology 19:1431-1440.

Owens, I. P. F., and P. M. Bennett. 2000. Ecological basis of extinction risk in

birds: Habitat loss versus human persecution and introduced predators.

Proceedings of the National Academy of Sciences of the United States

of America 97:12144-12148.

Pagel, M. 1993. Seeking the Evolutionary Regression Coefficient - an

Analysis of What Comparative Methods Measure. Journal of

Theoretical Biology 164:191-205.

Paradis, E., and J. Claude. 2002. Analysis of Comparative Data Using

Generalized Estimating Equations. Journal of Theoretical Biology

218:175-185.

Paradis, E., K. Strimmer, J. Claude, R. Opgen-Rhein, J. Dutheil, Y. Noel, and

B. Bolker. 2005. APE: analyses of phylogenetics and evolution., R

package v. 1.6. .

Parker, J. M., I. Mikaelian, N. Hana, and H. E. Diggs. 2002. Clinical diagnosis

and treatment of epidermal chytridiomycosis in African clawed frogs

(Xenopus tropicalis ). Comparative Medicine 52:265-268.

Pasmans, F., F. Mutschmanna, T. Halliday, and P. Zwart. 2007. Amphibian

decline: The urgent need for amphibian research in Europe. The

Veterinary Journal 171:18-19.

239 Pimm, S. L., H. L. Jones, and J. Diamond. 1988. On the risk of extinction. American Naturalist 132:757-785. Piotrowski, J. S., S. L. Annis, and J. E. Longcore. 2004. Physiology of

Batrachochytrium dendrobatidis , a chytrid pathogen of amphibians.

Mycologia 96:9-15.

Pounds, J. A. 2001. Climate and amphibian declines. Nature 410:639-640.

Pounds, J. A., M. R. Bustamante, L. A. Coloma, J. A. Consuegra, M. P. L.

Fogden, P. N. Foster, E. La Marca et al. 2006. Widespread amphibian

extinctions from epidemic disease driven by global warming. Nature

439:161-167.

—. 2007. Ecology - Pounds et al. reply. Nature 447:E5-E6.

Pounds, J. A., and M. L. Crump. 1994. Amphibian Declines and Climate

Disturbance: The Case of the Golden Toad and the Harlequin Frog.

Conservation Biology 8:72-85.

Pounds, J. A., M. P. L. Fogden, and J. H. Campbell. 1999. Biological

response to climate change on a tropical mountain. Nature 398:611-

615.

Puddu, F., M. Viarengo, and C. Ermineo. 1988. Euprotto sardo in D. Torre, ed.

Animali di Sardegna: gli Anfibi e i Rettili. Cagliari.

Purvis, A. 2008. Annual Review of Ecology and Evolution In press.

Purvis, A., P. M. Agapow, J. L. Gittleman, and G. M. Mace. 2000a.

Nonrandom extinction and the loss of evolutionary history. Science

288:328-330.

Purvis, A., M. Cardillo, R. Grenyer, and B. Collen. 2005a. Correlates of

extinction risk: phylogeny, biology, threat and scale. in A. Purvis, T. M.

Brooks, and J. L. Gittleman, eds. Phylogeny and Conservation.

240 Purvis, A., J. L. Gittleman, and T. M. i. p. Brooks. 2005b. Phylogeny and

Conservation in A. Purvis, T. M. Brooks, and J. L. Gittleman, eds.

Phylogeny and Conservation.

Purvis, A., J. L. Gittleman, G. Cowlishaw, and G. M. Mace. 2000b. Predicting

extinction risk in declining species. Proceedings of the Royal Society of

London Series B-Biological Sciences 267:1947-1952.

Purvis, A., and A. J. Webster. 1999. Phylogenetically independent

comparisons and primate phylogeny., Pages 44-70. In Comparative

primate socioecology (ed. P. C. Lee),. Cambridge, Cambridge

University Press.

Putland, D. 2005. Problems of studying extinction risks. Science 310:1277.

Quader, S., K. Isvaran, R. E. Hale, B. G. Miner, and N. E. Seavy. 2004.

Nonlinear relationships and phylogenetically independent contrasts

Journal of Evolutionary Biology 17:709-715.

R Development Core Team. 2005. R: A language and environment for

statistical computing. R Foundation for Statistical Computing,. Vienna,

Austria. .

Rachowicz, L. J., J.-M. Hero, R. A. Alford, J. W. Taylor, J. A. T. Morgan, V. T.

Vredenburg, J. P. Collins et al. 2005. The Novel and Endemic

Pathogen Hypotheses: Competing Explanations for the Origin of

Emerging Infectious Diseases of Wildlife. Conservation Biology

19:1441-1448.

Rachowicz, L. J., R. A. Knapp, J. A. T. Morgan, M. J. Stice, V. T. Vredenburg,

J. M. Parker, and C. J. Briggs. 2006. Emerging infectious disease as a

proximate cause of amphibian mass mortality. Ecology 87:1671-1683.

241 Rachowicz, L. J., and V. T. Vredenburg. 2004. Transmission of

Batrachochytrium dendrobatidis within and between amphibian life

stages. Diseases of Aquatic Organisms 61:75-83.

Read, A. F., and S. Nee. 1995. Inference from binary comparative data.

Journal of Theoretical Biology 173:99-108.

Read, A. W. 1998. On Euproctus platycephalus . British Herpetological Society

Bulletin 62.

Redding, D. W., and A. O. Mooers. 2006. Incorporating evolutionary

measures into conservation prioritization. Conservation Biology

20:1670-1678.

Reed, R. N., and R. Shine. 2002. Lying in Wait for Extinction: Ecological

Correlates of Conservation Status among Australian Elapid Snakes.

Conservation Biology 16:451-461.

Retallick, R.W.R, McCallum, H., and R.Speare. 2004. Endemic Infection of

the Amphibian Chytrid Fungus in a Frog Community Post-Decline.

PLoS Biology 2(11) e351. doi:10.1371/journal.pbio.0020351

Ricketts, T. H., E. Dinerstein, T. Boucher, T. M. Brooks, S. H. M. Butchart, M.

Hoffmann, J. F. Lamoreux et al. 2005. Pinpointing and preventing

imminent extinctions. Proceedings of the National Academy of

Sciences of the United States of America 102:18497-18501.

Ricklefs, R. E., and J. M. Starck. 1996. Applications of phylogenetically

independent contrasts: A mixed progress report. Oikos 77:167-172.

Ridley, M. 1983, The explanation of organic diversity: the comparative method

and adaptations for mating. Oxford, Oxford University Press.

242 Rodrigues, A. S. L., J. D. Pilgrim, J. F. Lamoreux, M. Hoffmann, and T. M.

Brooks. 2006. The value of the IUCN Red List for conservation. Trends

in Ecology & Evolution 21:71-76.

Ron, S. 2005. Predicting the Distribution of the Amphibian Pathogen

Batrachochytrium dendrobatidis in the New World. Biotopica 37:209-

221.

Russell, G. J., T. M. Brooks, M. M. McKinney, and C. G. Anderson. 1998.

Present and future taxonomic selectivity in bird and mammal

extinctions. Conservation Biology 12:1365-1376.

Schenk, H., M. Aresu, and A. Fozzi. 1995, Libro rosso dei vertebrati terrestri

del Marghine-Planargia. Legambiente- Circolo di iniziativa ambientale

(Macomer).

Schloegel, L. M., J. M. Hero, L. Berger, R. Speare, K. McDonald, and P.

Daszak. 2006. The decline of the sharp-snouted day frog ( Taudactylus

acutirostris ): The first documented case of extinction by infection in a

free-ranging wildlife species? Ecohealth 3:35-40.

Sechrest, W., T. M. Brooks, G. A. B. da Fonseca, W. R. Konstant, R. A.

Mittermeier, A. Purvis, A. B. Rylands et al. 2002. Hotspots and the

conservation of evolutionary history. Proceedings of the National

Academy of Sciences of the United States of America 99:2067-2071.

Seddon, P. J., P. S. Soorae, and F. Launay. 2005. Taxonomic bias in

reintroduction projects. Animal Conservation 8:51-58.

Simoncelli, F., A. Fagotti, R. Dall’Olio, D. Vagnetti, R. Pascolini, and I. Di

Rosa. 2005. Evidence of Batrachochytrium dendrobatidis Infection in

243 Water Frogs of the Rana esculenta Complex in Central Italy Ecohealth

2:307-312.

Sodhi, N. S., D. Bickford, A. C. Diesmos, Lee T.M., Koh L.P., B. W. Brook, C.

H. Sekercioglu et al. 2008. Measuring the Meltdown: Drivers of Global

Amphibian Extinction and Decline. . PLoS ONE 3.

Sokal, R. R., and F. J. Rolf. 1995, Biometry: The Principles and Practice of

Statistics in Biological Research. New York, W.H. Freeman.

Soto-Azat, C. 2007. Out of Africa? On the origin of amphibian

chytridiomycosis. Master of Science in Wild Animal Health thesis,

University of London.

Stagni, G., C. Scoccianti, and R. Fusini. 2002. Segnalazione di chytridiomicosi

in popolazioni di Bombina pachypus (Anura, Bombinatoridae)

dell’Appennino toscoemiliano., Abstracts IV; Congresso dellaSocietas

Herpetologica Italica. Napoli, Societas Herpetologica Italica.

Stuart-Fox, D., A. Moussalli, and M. J. Whiting. 2007. Natural selection on

social signals: Signal efficacy and the evolution of chameleon display

coloration. American Naturalist 170:916-930.

Stuart, S. N., J. S. Chanson, N. A. Cox, B. E. Young, A. S. L. Rodrigues, D. L.

Fischman, and R. W. Waller. 2004. Status and Trends of Amphibian

Declines and Extinctions Worldwide. Science 306:1783-1786.

Sullivan, M., M. Jones, D. C. Lee, S. J. Marsden, A. H. Fielding, and E. V.

Young. 2006. A comparison of predictive methods in extinction risk

studies: contrasts and decision trees. Biological Conservation 15:1977-

1991.

244 Sullivan, M. S., F. Gilbert, G. Rotheray, S. Croasdale, and M. Jones. 2000.

Comparative analyses of correlates of Red data book status: a case

study using European hoverflies (Diptera:Syrphidae). . Animal

Conservation 3:91-95.

Sutherland, W. J. 2006. Predicting the ecological consequences of

environmental change: a review of the methods. Journal of Applied

Ecology 43:599-616.

Sutherland, W. J., A. S. Pullin, P. M. Dolman, and T. M. Knight. 2004. The

need for evidence-based conservation. Trends in Ecology & Evolution

19:305-308.

Symonds, M. R. E. 2002. The Effects of Topological Inaccuracy in

Evolutionary Trees on the Phylogenetic Comparative. Systematic

Biology 51:541-555.

Thomas, C. D., and J. C. G. Abery. 1995. Estimating rates of butterfly decline

from distribution maps: the effect of scale. Biological Conservation

73:59-65.

Thrusfield, M. 2005, Veterinary Epidemiology, Blackwell Science Ltd.

Van Rooy, T. J. C., and A. H. P. Stumpel. 1995. Ecological impact of

economic development on Sardinian herpetofauna. Conservation

Biology 9:263-269.

Vilella, F. J., and P. J. Swank. 1993. Ecology of the small indian mongoose in

a coastal dry forest of Puerto Rico where sympatric with the Puerto

Rican nightjar. Caribbean Journal of Science 29:24-29.

Wake, D. B. 1991. Declining Amphibian Populations. Science 253:860-860.

245 Walker, S., M. B. Salas, D. Jenkins, T. W. J. Garner, A. A. Cunningham, A.

Hyatt, J. Bosch et al. 2007. Environmental Detection of

Batrachochytrium dendrobatidis in a temperate climate. Diseases of

Aquatic Organisms 77:105-112.

Weldon, C., and L. H. du Preez. 2004. Decline of the Kihansi Spray Toad,

Nectophrynoides asperginis , from the Udzunga Mountains, Tanzania.

Froglog 62.

Weldon, C., L. H. du Preez, A. D. Hyatt, R. Muller, and R. Speare. 2004.

Origin of the amphibian chytrid fungus. Emerging Infectious Diseases

10:2100-2105.

Whitfield, S. M., K. E. Bell, T. Philippi, M. Sasa, F. Bolanos, G. Chaves, J. M.

Savage et al. 2007. Amphibian and reptile declines over 35 years at La

Selva, Costa Rica. Proceedings of the National Academy of Sciences

of the United States of America 104:8352:8356.

Williams, S. E., and J. M. Hero. 1998. Rainforest frogs of the Australian Wet

Tropics: guild classification and the ecological similarity of declining

species. Proceedings of the Royal Society of London Series B-

Biological Sciences 265:597-602.

Willmott, C. J., and Matsuura Kenji. 2001. Terrestrial Water Budget Data

Archive: Monthly Time Series (1950-1999). .

Woodhams, D. C., K. Ardipradja, R. A. Alford, G. Marantelli, L. K. Reinert, and

L. A. Rollins-Smith. 2007. Resistance to chytridiomycosis varies among

amphibian species and is correlated with skin peptide defenses. Animal

Conservation 10:409-417.

246 Wyman, R. L. 1990. Whats Happening to the Amphibians. Conservation

Biology 4:350-352.

Young, B. E., K. R. Lips, J. K. Reaser, R. Ibanez, A. W. Salas, J. R. Cedeno,

L. A. Coloma et al. 2001. Population Declines and Priorities for

Amphibian Conservation in Latin America. Conservation Biology

15:1213-1223.

247