CONSERVATION OF PLANT SPECIES UNDER FUTURE CLIMATE AND LAND-USE CHANGE GIAM XINGLI [B.Sc.(Hons.), National University of Singapore] A THESIS SUBMITTED FOR THE DEGREE OF MASTER OF SCIENCE DEPARTMENT OF BIOLOGICAL SCIENCES NATIONAL UNIVERSITY OF SINGAPORE 2009 i ACKNOWLEDGEMENTS I would like to thank the Department of Biological Sciences, National University for funding my research through the graduate research scholarship. Additional support was provided in the form of the Endeavour Research Fellowship by the Government of Australia to fund my research visit to The Environment Institute of The University of Adelaide. I would like to express my sincerest gratitude to my supervisors, Associate Professor Hugh Tan, Professor Navjot Sodhi, and Associate Professor Corey Bradshaw (The University of Adelaide) for their invaluable guidance, insightful comments, and ready support. My three supervisors have really helped me to grow as an ecologist-intraining, as well as a person, in the last two to three years. I also thank the two examiners for critical comments that helped to improve the thesis. I would also like to thank Professor Barry Brook (The University of Adelaide) for insightful discussions and commenting on parts of my thesis. Also, I am grateful to Tien Ming Lee (University of California, San Diego) for readily supplying raw data for future land transformation, helpful discussions, and commenting on parts of my thesis. My colleagues in the Plant Systematics Lab provided me with the motivation to turn up for work every morning with their wicked sense of humour, free coffee, biscuits, home-baked cakes, and most importantly, friendship. I thank Ang Wee Foong, Chong Kwek Yan, Alvin Lok, Ng Peixin, Ng Ting Hui, Edwin Phua, Tan Kai-xin, Alex Yee, and Yeo Chow Khoon for that. I am grateful to the members of the Global Ecology Laboratory for helpful discussions and making me feel at home in my five month research stint at The University of Adelaide. I thank Bert Harris, Salvador Herrando-Perez, Siobhan de Little, Tom Wanger, Ana Sequiera, Michael Stead, Zheng Dandong, Drs. Damien Fordham and Camille Mellin. In particular, Drs. Steven Delean and Lochran Traill provided insightful comments and helpful advice that improved the thesis. Finally, my deepest appreciation and thanks to my family and my lovely wife, Gillian Goh-Giam. Their unflinching support and unconditional love made my life more meaningful and gave me the added strength and motivation needed for my endeavours. i TABLE OF CONTENTS Acknowledgements i Table of contents ii Summary iv List of tables vi List of figures vii List of appendices viii 1. INTRODUCTION 1 2. FUTURE HABITAT LOSS AND THREATENED PLANT BIODIVERSITY 2.1. INTRODUCTION 5 2.2. METHODS 2.2.1. Number of globally threatened plant species 8 2.2.2. National index of relative plant species endangerment 10 2.2.3. Future habitat loss from land-use and climate changes 11 2.2.4. Future plant species endangerment rankings 12 2.2.5. Current effort in conducting conservation assessments of plant species by IUCN 13 2.2.6. Wealth, governance, and future conservation need 13 2.3.1. Patterns of current plant species endangerment 16 2.3.2. Exposure of threatened plant biodiversity to future habitat loss 18 2.3.2.1. Habitat loss owing to climate change 18 2.3.2.2. Habitat loss owing to land-use change 20 2.3. RESULTS 2.4. 2.3.3. Future plant species endangerment rankings 22 2.3.4. Countries with the greatest future conservation need 24 DISCUSSION 26 ii 3. RELATIVE NEED FOR CONSERVATION ASSESSMENTS OF PLANT SPECIES AMONG ECOREGIONS 3.1. INTRODUCTION 33 3.2. METHODS 3.2.1. Ranking ecoregions based on plant species richness 36 3.2.2 . Historical habitat loss 40 3.2.3. Future human population pressure (2005-2015) 40 3.2.4. Allocating the relative need for conservation assessments among ecoregions 40 3.2.5. Testing the overlap of this template with existing schemes 41 3.2.6. Availability of financial resources in ecoregions with high relative need for conservation assessments 42 3.3. RESULTS 3.3.1. Ecoregions of high relative need for conservation assessments 42 3.3.2. Species-area models used in ranking plant species richness 46 3.3.3. Overlap with existing schemes 51 3.3.4. Lack of financial resources in ecoregions important for conservation assessments 52 3.4. DISCUSSION 55 4. CONCLUSIONS 58 5. REFERENCES 61 6. APPENDICES 71 iii SUMMARY Habitat loss resulting from human land-use change and climate change threatens plant biodiversity on Earth. I assessed the vulnerability of globally threatened plant biodiversity to future habitat loss over the first half of this century by testing countrylevel associations between threatened plant species richness and future habitat loss owing to, separately, land-use and climate changes. In countries that overlap with biodiversity hotspots, plant species endangerment increases with climate change-driven habitat loss. The same pattern was observed among tropical countries. This association suggests that many currently threatened plant species may become extinct owing to anthropogenic climate change in the absence of potentially mitigating factors such as natural and assisted range shift, and physiological and genetic adaptations. Countries ranked the highest in future plant species endangerment are concentrated around the equator. The current conservation assessment effort by IUCN was positively correlated with future plant species endangerment, suggesting that the conservation assessment program is generally efficient in targeting the most threatened countries. Because poverty and poor governance can compromise conservation, I considered the economic condition and quality of governance with future plant species endangerment to prioritize countries based on conservation need. I identified Angola, Cuba, Democratic Republic of Congo, Ethiopia, Kenya, Laos, Madagascar, Myanmar, Nepal, Tajikistan, and Tanzania as the countries in greatest need of conservation assistance in terms of financial aid and/or improving political institutions. Conservation assessments aid in planning by providing valuable information about the geographic range and population numbers of species. However, less than 5% of all iv plant species have been assessed. I therefore aimed to provide a template to guide conservation assessments at the ecoregion level. First, I identified the world‘s ecoregions that contain the highest plant species richness after controlling for area using species-area relationship (SAR) models within a Bayesian multi-model framework. While all previous studies have assumed that species richness is normally distributed and most applied the power function SAR, I found that species richness was log-normally distributed across ecoregions in most biomes and no SAR model was the best in all biomes. My results highlight the importance of considering a wide variety of SAR models with different error distributions to identify species-rich hotspots. Using quantitative thresholds, ecoregions with the highest plant species richness, historical habitat loss, and projected increase in human population density were allocated the highest relative need for conservation assessments. My template managed to identify some important ecoregions excluded from the Global 200 and Biodiversity Hotspots templates. Using generalized linear models, I showed that countries overlapping with high-priority ecoregions are poorer than the other countries. Therefore I urge international aid agencies and botanic gardens to cooperate with local scientists to fund and implement conservation assessment programs. Overall, my study showed that plant biodiversity remains vulnerable to climate change driven habitat loss, and socioeconomic factors. The international community must consider both global and local strategies that aim to improve governance and economic condition for conservation endeavours to be truly effective. v LIST OF TABLES Table 1. Total variance explained by each component of the 14 Principal Components Analysis (PCA) of governance quality. Table 2. Correlations between the six dimensions of governance 14 and the principal component. Table 3. Species-area relationship (SAR) models investigated in this study. 38 Table 4. DIC weights for species-area relationship models 46 fitted to each each biome. Table 5. List of 21 ecoregions excluded from both Biodiversity Hotspots 52 and Global 200 prioritization template but included in my prioritization template (40th percentile threshold). Table 6. Generalized linear models (GLM) investigating the correlation of per capita wealth on whether a country overlaps with important ecoregions selected using multiple percentile thresholds (a–f). vi 54 LIST OF FIGURES Figure 1. Global map showing the current plant endangerment index of 17 (a) all countries, (b) hotspot countries, (c) non-hotspot countries, (d) tropical countries, and (d) temperate countries Figure 2. Relationship between the ranks of current endangerment index 19 and ranks of future habitat loss owing to climate change. Figure 3. Relationship between the ranks of current endangerment index 21 and ranks of future habitat loss owing to land-use change. Figure 4. Global map of the country rankings for future plant 23 species endangerment. Figure 5. Countries classed into five categories of conservation need 25 (Categories 1–5) in terms of (a) economic wealth, and (b) governance quality. Figure 6. Relative need for the conservation assessment of plant species among 756 global ecoregions. 45 Figure 7. Individual and model-averaged special-area relationship 47 (SAR) models fitted using a Markov Chain Monte-Carlo (MCMC) procedure implemented in WinBUGS. vii LIST OF APPENDICES Appendix 1. Additional sources of total and endemic species data used in analyses. 71 Appendix 2. Supplementary tables 74 Table S1. Generalized linear models (GLM) investigating whether an increase in area decreases the odds of endangerment given that odds of endemism are kept constant. 74 Table S2. List of countries ranked by the current endangerment index of plant species. 75 Table S3. Spearman’s rank-order correlations between current plant species endangerment index and future habitat loss. 81 Table S4. List of countries ranked in terms of conservation need. 82 Table S5. Full dataset used in allocating the relative conservation assessment need of ecoregions. 83 Table S6. List of ecoregions and their respective overlapping biodiversity hotspots and Global 200 ecoregions. 110 Table S7. Countries overlapping with important ecoregions at multiple thresholds (10th to 80th percentile). 132 Appendix 3. Supplementary figures 138 Figure S1. Global distribution of habitat loss owing to climate change and land-use change. 138 Figure S2. Fitted curve, normal probability plot, and residual plot of non-linear species-area relationship (SAR) models, fitted by mimimizing the residual sum of squares. 139 Figure S3. The number of ecoregions and their breakdown by biome type in each category of relative conservation assessment need. 153 Figure S4. The number of ecoregions and their breakdown by biogeographic realm type in each category of relative 154 viii conservation assessment need. Figure S3. Location of the two ecoregions excluded from both G200 and BH templates. Appendix 4. Statistical codes for WinBUGS ix 155 157 1. INTRODUCTION Plant species, being autotrophs, are the fundamental components of most ecosystems on Earth. They support non-plant taxa by serving as the foundation of most food webs (Huston 1994; Primack & Corlett 2005), and are involved in many ecological processes necessary for the maintenance of life on Earth (Hamilton & Hamilton 2006). They also provide food and materials for humans (Kier et al. 2005). However, plant species and other terrestrial biodiversity are endangered by habitat losses resulting from anthropogenic land-use changes and climatic changes (Millennium Ecosystem Assessment 2005; Thomas et al. 2006; Bradshaw et al. 2009a). In the tropics, there is convincing evidence of rapid forest decline in recent decades (Bradshaw et al. 2009a) of 10.2 million ha annually (Hansen & DeFries 2004). Temperate grasslands, temperate broadleaf forests, and Mediterranean forests have all experienced at least 35% conversion to cropland (Millennium Ecosystem Assessment 2005), and even the boreal forest zone has become highly fragmented (Bradshaw et al., 2009b). Recent climate warming has already resulted in geographical range contraction of butterflies (Wilson et al. 2005). Land transformations resulting from climate change and land use change are projected to cause a 21-26% reduction in the mean geographic range of bird species by year 2050 (Jetz et al. 2007). With high expected rates of continued habitat loss owing to anthropogenic activities, many plant species face extinction, thereby compromising ecosystem services that sustain the quality of life for billions of people (Daily 1997; Ehrlich & Pringle 2008). This begs the question, what conservation strategies can we employ to prevent or slow 1 down the loss of plant species? This thesis, consisting of two main papers (Chapters 2 and 3), aims to answer this main question as well as other related questions. In Chapter 2 (Future Habitat Loss and Threatened Plant Biodiversity1), I assessed the exposure of threatened plant biodiversity to land-use and climate change-driven habitat loss by testing the country-level correlation between the level of plant species endangerment and climate change induced habitat loss, as well as human land-use change induced habitat loss, separately. Next, I related the current species endangerment level to future habitat loss to determine the future level of endangerment following the ranking procedures of Lee & Jetz (2008). To evaluate the efficacy of the current prioritization pattern of IUCN species assessments under future scenarios, I tested the association between the proportion of species assessed in a country and future plant species endangerment. Lastly, I incorporated economic condition and quality of governance with future plant species endangerment ranks to determine countries of high conservation need. By considering these projected impacts together with governance quality and poverty, my system of conservation ranking allows national lawmakers and the international community to prioritize conservation management. In Chapter 3 (Relative need for conservation assessments of plant species among ecoregions2), I provide a new template to guide the conservation assessment of plant species using the World Wildlife Fund (WWF) ecoregions framework (Olson et al. 2001). Conservation assessments are conducted to acquire knowledge on the population numbers and geographic range of all species, so that conservation managers can focus on 1 The contents of this chapter have been incorporated into a paper submitted to the journal, Conservation Biology. The paper has been reviewed by two anonymous referees and is currently in revision stage. 2 The contents of this chapter have been incorporated into a paper submitted to the journal, Journal of Biogeography. The paper is currently in review. 2 species with greatest risk of extinction. However conservation assessments have only been carried out on less than 5% of all plant species, hence many plant species may be driven to extinction even before their population numbers and range become known to scientists. Speeding up the assessment process is therefore crucial to effective conservation. However, economic and other resources available for conservation are finite, and especially limited in the developing world (Sodhi et al. 2007), which mandates the use of prioritization. Species-area curves are commonly used to identify the most speciose ecoregions (Fattorini 2006; Guilhaumon et al. 2008) and I demonstrated the importance of, and provided a framework for, considering uncertainty in the species-area relationship (SAR) models and distribution. Ecoregions with the highest species richness, and at the same time, highest degree of habitat loss and future human population pressure, is likely to have the greatest number of declining and were hence allocated the highest need for conservation assessments. Lastly, to evaluate the challenge posed by the lack of financial resources, I used generalized linear models (GLM) to test if countries that overlap with ecoregions deemed important for conservation assessments are poorer compared to others. In this thesis, I provided the first global-scale assessment of the association between threatened plant species and future land use- and climate change-driven habitat loss. My study is the first attempt to prioritize the conservation assessment of plant species by taking into account uncertainty in the form and distribution of SAR models, as well as, current and future endangerment levels by introducing historical habitat loss and future population change as criteria for prioritization. I hope that the results of this study translate to the intensification of conservation effort in countries with the greatest 3 conservation need and stimulate increased scientific activity that contributes to the conservation assessment of plant species in specious, and at the same time, highlythreatened ecoregions. 4 2. FUTURE HABITAT LOSS AND THREATENED PLANT BIODIVERSITY 2.1. INTRODUCTION Human-driven land-use and climatic changes are perhaps the greatest threats to terrestrial biodiversity (Millennium Ecosystem Assessment 2005a; IPCC 2007) given the mounting empirical evidence that these anthropogenic forcings substantially exacerbate species‘ endangerment (Brook et al. 2003; Sodhi et al. 2008). As these environmental changes are likely to continue into the future (Millennium Ecosystem Assessment 2005b), it is important to assess the impacts of these changes on biodiversity for effective prioritization of future conservation efforts (Lee & Jetz 2008). In particular, the impacts of land-use and climatic changes on plant biodiversity will have extensive ramifications on other taxa and human society given that plants are fundamental structural and nutrient-sequestering components of most ecosystems. Not only do plants produce resources that support non-plant biodiversity (Huston 1994; Primack & Corlett 2005), they also provide food and materials essential for human existence (Kier et al. 2005), and are involved in many ecological processes necessary for the persistence of life (Hamilton & Hamilton 2006). Several studies have predicted the future extinction patterns of plant species based on land-use and climate change projections (Thuiller et al. 2005; Van Vuuren et al. 2006), but none has explicitly examined the association between the current endangerment and future habitat loss (e.g., Lee & Jetz 2008 for vertebrates). Threatened plant species are more likely to be driven towards extinction by future habitat loss than non-threatened species because the former are restricted in range and consist of fewer individuals. I can therefore determine whether 5 future land-use and climatic changes exacerbate extinction risk predictions by testing whether these will have the greatest negative influences in areas already characterized the highest number of currently threatened species. Therefore, realistic projections of conservation impact should not only take into account the magnitude of predicted habitat loss and degradation, they must also take into the account the distribution of species currently threatened with extinction. Here I assessed the exposure of threatened plant biodiversity to land-use and climate change-driven habitat loss up to year 2050 by testing the hypothesis that countries with more threatened plant species richness (after controlling for the effect of area) are likely to suffer from greater relative habitat loss given recent historical trends. I estimated country-specific plant species endangerment by (i) using the number of endemic plant species per country as a proxy for the number of threatened species (see Methods and Supplementary Methods, Online Supplementary Material [SM]), and (ii) using the residuals in the power-law endemic species-country area relationship as an index of endangerment. I then quantified the potential extent of future habitat loss owing to land-use and climate changes up to 2050 in each country (data from Lee & Jetz 2008) derived from the Millennium Ecosystem Assessment (2005b). I elucidated the distribution of future plant species endangerment by relating the current level of plant species endangerment with the degree of projected total habitat loss (Lee & Jetz 2008) to determine which countries are most prone to plant biodiversity loss. To evaluate the efficacy of the current prioritization pattern of IUCN species assessments under future scenarios, I tested the association between the proportion of species assessed in a country and future plant species endangerment. I considered countries that overlap with 6 biodiversity hotspots (Myers 2000; Mittermeier et al. 2004) separately to those that do not because biodiversity hotspots are considered urgent conservation priorities owing to high plant endemicity (~150000 endemic species in total) and high historical habitat loss (Myers 2000; Mittermeier et al. 2004). Like biodiversity hotspots, tropical regions are focal areas for conservation because they are highly biodiverse, but at the same time, threatened by high rates of habitat degradation (Laurance 2007; Bradshaw et al. 2009a). I therefore considered tropical countries separately from temperate countries in my analyses. Less wealthy countries have limited financial means for conservation projects such the enforcement and monitoring of protected areas (Bruner et al. 2004); hence, species there may be at greater risk from habitat loss, direct harvesting, and encroachment of invasive alien species. Poverty may also lead to unsustainable exploitation of resources (Kerr & Currie 1995) and could therefore exacerbate species loss through direct harvesting. Poor enforcement of existing legislation, weak governance and lack of political will and corruption can result in the degradation of biodiversity owing to ineffective conservation management (O'Connor et al. 2003) and high deforestation rates in developing countries (Geist & Lambin 2002). In addition, conservation efforts may be compromised by decision-making processes in other sectors (Deutz 2005), such as, economic planning and residential land-use planning, in the absence of effective high-level coordination within and between national ministries (Bojö & Chandra Reddy 2001). As poverty and poor governance were shown to have adverse impacts on biodiversity conservation, I identified countries of high conservation need by considering their wealth and quality of governance with the future plant species 7 endangerment ranks. Poor countries with low-quality governance and high future plant species endangerment were identified as having the highest conservation need. I provided the first global-scale assessment of the association between threatened plant species and future land use- and climate change-driven habitat loss, and present plausible policies towards plant species conservation. By considering these projected impacts together with governance quality and poverty, my system of conservation ranking allows national lawmakers and the international community to prioritize conservation management. 2.2. METHODS 2.2.1. Number of globally threatened plant species The number of endemic species was used as a proxy for the number of threatened species because the large taxonomic gap in the current (post-1997) IUCN Red List ― where only about 12000 species out of a total flora of 223300–422000 species were assessed ― limits its use to infer global patterns of extinction risk (Pitman & Jorgensen 2002). My two main sources of country-level data on the number of endemic plant species were datasets from Pitman & Jorgensen (2002) and United Nations Environment Programme‘s World Conservation Monitoring Centre (UNEP-WCMC) (World Resources Institute 2007). For countries with missing data, I used values in national biodiversity reports and Floras wherever possible (Appendix 1). Although the number of endemic species is correlated with the number of threatened species (Pearson's r = 0.78, in European countries with reliable threatened species data, Pitman & Jorgensen 2002), using endemism as a proxy for endangerment 8 may be confounded by differences in the size of each country. For instance, it is logical to posit that species endemic to a large country are less likely to be endangered because their potential range size is larger than that of species endemic to a smaller country. To test for this possible bias, I constructed a set of generalized linear models (GLM) with a binomial error distribution and logit link function. I set the log-proportional odds of endangerment [loge(number of threatened species ÷ number of non-threatened species)] as the response and the loge-transformed country area and the log-proportional odds of endemism [loge(number of endemic species ÷ number of non-endemic species)] as the predictors. I computed a measure of overdispersion by dividing the observed deviance of the global model by its degrees of freedom (Burnham & Anderson 2002; Franklin et al. 2002). Following Pitman and Jorgensen (2002), I used data from European countries only (n = 32) because temperate countries tend to have more accurate tallies of threatened flora than tropical countries. I followed Pitman & Jorgensen (2002) in using the number of globally threatened species listed in the 1997 IUCN Red List of Threatened Plants (Walter & Gillett 1998) because the 1997 Red List evaluated the threat levels of most species on Earth by combining national and regional threatened species lists, while only about 12000 out of a total global flora of about 223300–422000 species (Prance et al., 2000; Govaerts, 2001; Bramwell, 2002; Scotland & Wortley, 2003) were assessed up to year 2008 in all post-1997 IUCN Red Lists. The most parsimonious model evaluated by QAICc (quasi Akaike‘s information criterion corrected for small sample sizes and over-dispersion) contained both log-odds of endemism and log-transformed area as predictors of the log-transformed odds of endangerment (wQAICc [model weight] = 0.675) (Table S1 in Appendix 2). This model 9 refutes the suggested bias because the odds of endangerment increase, rather than decrease, with area when odds of endemism are kept constant. These results suggest that the potential bias due to country area is absent or weak. The second-ranked model in which log-odds of endangerment increases with the log-odds of endemism (wQAICc =0.396) had a high percentage deviance explained (~ 41%), suggesting the level of endemism is the main predictor of endangerment, not area. I therefore argue that the number of endemic species is a good proxy for the number of threatened species when accurate endangerment data are absent across countries that overlap with biodiversity hotspots. 2.2.2. National index of relative plant species endangerment Among-country comparisons of the raw number of globally threatened species cannot be made because of differing land areas; therefore, I fitted the power-law species-area relationship (SAR) (S = cAz, where S = endemic species richness as a proxy for number of threatened species, A = country area, z = the power coefficient and c = a constant; Arrhenius 1921) to 196 countries and considered the residuals as a proxy of relative species endangerment. Country-area data were obtained from the World Resources Institute (2007) EarthTrends database. I am cognizant that the curvilinear form of the SAR is likely to provide a more realistic detection of the hotspots of endangerment compared to the linearized form (Fattorini 2007); hence, I first fitted the curvilinear form of the power-law SAR using the nls function in R v.2.8 (R Development Core Team 2008). I calculated starting parameter values based on standard procedures described in Ratkowsky (1990). However, the curvilinear model was untenable because the residuals 10 were ascertained to be non-Gaussian and heteroscedastic via visual inspection of the residual plot. I then reverted to fit the linearized SAR model (loge[number of endemic species + 1] ~ loge[country area]) (e.g., Balmford & Long 1995; Lee & Jetz 2008) using glm in R. Log-transformed area explained ~ 25.5 % of the deviance in the logtransformed number of endemic species. The residuals of this model were taken as an index of plant species endangerment controlling for country area. 2.2.3. Future habitat loss from land-use and climate changes The Millennium Ecosystem Assessment (2005a) developed four socioeconomic scenarios that delineated possible future outcomes of terrestrial ecosystem change up to 2050 (Adapting Mosaic [AM], Global Orchestration [GO], Order from Strength [OS], and TechnoGarden [TG]). Land-cover projections in the MEA were made based on the IMAGE v. 2.2 model (Image-Team 2001), which provided current and projected areal distributions for 18 land-cover classes at 0.5° resolution. The IMAGE model generates explicit forecasts of land cover using a set of linked and integrated socioeconomic, climate and environmental models (described in detail in Alcamo et al. 1998; Millennium Ecosystem Assessment 2005b). I obtained raw data from Lee and Jetz (2008) who calculated the percentage of area subjected to land-cover transformation in 174 countries owing to land-use and climate changes over the four MEA socioeconomic scenarios from 2000 to 2050. Lee and Jetz (2008) classified transformation from natural- to humaninduced land cover types as land-use-driven (e.g., pristine forest converted to agricultural land), and change from one natural land-cover category (e.g., pristine forest converted to savanna) as driven by climate change. Thus, my metric for climate change-driven habitat 11 loss was the average area projected to undergo transformation from one natural landcover category to another under the four socioeconomic scenarios to the year 2050 expressed as a percentage of the total land area of a country, and the metric for future habitat loss owing to land-use change was the average area projected to be converted via human land-use change divided under the four scenarios to year 2050 expressed as a percentage of total land area. I used the percentage land-cover transformations averaged over all four scenarios for my main analyses because the ‗actual‘ future is likely to fall in between the four scenarios (Millennium Ecosystem Assessment 2005b). I also conducted additional analyses using land-cover transformation data from each of the four scenarios to examine the sensitivity of my results towards the dominance of any one particular scenario. I used Spearman‘s ρ rank-order correlation to test the correlation between the index of species endangerment and future habitat loss owing to land-use and climate changes. 2.2.4 Future plant species endangerment rankings By relating the current level of plant species endangerment to total future habitat loss, I ranked 163 countries (with data available for future habitat loss and current endangerment index) according to their future level of plant species endangerment (Lee & Jetz 2008). Each country was separately ranked in terms of total future habitat loss (from both land-use and climate changes) and current species endangerment. I averaged the percentile rank values of these two measures and re-ranked the derived value in descending order to obtain a global ranking of future plant species endangerment (after Lee & Jetz 2008). 12 2.2.5. Current effort in conducting conservation assessments of plant species by IUCN The IUCN Red List categorizes species into relative threat categories and provides information on the reasons for the categorization. To quantify current effort in assessing species endangerment, I calculated the proportion of the number of plant species assessed in each country up to year 2008 in the current (post-1997) Red List (www.iucnredlist.org). The total number of plant species in each country was the average of the numbers collated from Pitman and Jorgensen (2002), and UNEP-WCMC (World Resources Institute 2007). I used Spearman‘s ρ rank-order correlation to test the concordance between the current effort in assessing species endangerment and future plant species endangerment. 2.2.6. Wealth, governance, and future conservation need I adopted per capita gross national income adjusted for purchasing power parity (GNIPPP) averaged from 2003 to 2007 as a measure of a country‘s relative wealth. Per capita GNI-PPP data were collated from the World Bank World Development Indicators database (www.worldbank.org/data). I obtained governance quality data from the 2008 Worldwide Governance Indicators (WGI) project (Kaufmann et al. 2008) that appraised countries using indicators of six dimensions of governance: voice and accountability, political stability, government effectiveness, regulatory quality, rule of law, and control of corruption. For each of the six indicators, a score of –2.5 (lowest quality of governance) to 2.5 (highest quality of governance) was allocated to each country. 13 Because future projections of governance were not available, I calculated average values of each of these six dimensions for each country from 2003 to 2007 to obtain a plausible estimate of the relative future governance quality of each country. I used principal component analysis to extract only one component (explaining 86.8 % of the variance; Table 1) consisting of all six dimensions because of strong inter-correlations (Table 2). Human population increase (Davies et al. 2006) was excluded in my analyses for conservation need because it was used to model land-use change in IMAGE 2.2. Table 1. Total variance explained by each component of the Principal Components Analysis (PCA) of governance quality. Only one component (1) is extracted. Initial Eigenvalues Component Total % of Cumulative Variance % 86.788 86.788 1 5.207 2 .375 6.253 93.042 3 .262 4.360 97.402 4 .091 1.509 98.911 5 .041 .682 99.593 6 .024 .407 100.000 Extraction Sums of Squared Loadings % of Cumulative Total Variance % 5.207 86.788 86.788 Table 2. Correlations between the six dimensions of governance and the principal component Component 1 Dimensions of governance Correlations .885 Voice and accountability Political stability .842 Government effectiveness .965 Regulatory quality .952 14 Rule of law .974 Control of corruption .964 By relating the future plant species endangerment (see section 2.2.4) to economic wealth and governance quality, I took into account the exacerbating effects of poverty and poor governance on biodiversity and identifed countries with the greatest need for conservation efforts. My analysis of future conservation need included 145 countries (using economic wealth as a metric) and 157 (using governance quality as a metric) countries after removing 18 countries with unavailable wealth data and six countries with unavailable governance quality data. Like previous studies (e.g., Myers et al. 2000), I introduced quantitative thresholds for the designation of high-priority countries. This multiple-threshold method ranked countries in five categories of decreasing future conservation need. I assigned countries ranked in the top 20 % in future plant species endangerment, and the bottom 20 % in governance quality or per capita GNI-PPP, as those having the greatest conservation need (Category 1). I placed countries ranked in successive 20% increments in future plant species endangerment and governance quality or wealth in categories of decreasing conservation need (e.g., 60%, Category 2; 40%; Category 3). I recognize that the economic condition and governance quality of countries may change quickly especially in politically-turbulent regions (e.g., sub-Saharan Africa) and may affect the accuracy of my analyses. However, as the future economic condition and governance quality cannot be quantified with certainty, and that the effects of economic condition and governance quality on species endangerment is currently on-going, I argue 15 that there is adequate merit in using present-day wealth and quality of governance data to guide future conservation efforts. 2.3. RESULTS 2.3.1. Patterns of current plant species endangerment Endemic species richness data were available for 196 countries. One hundred forty-three countries overlap with the network of biodiversity hotspots and they contain a total of 206905 endemic plant species which represent ~ 64.1 % of the global flora based on the mean of the estimated lower (223300; Scotland & Wortley 2003) and upper limits (422000; Govaerts 2001; Bramwell 2002). In contrast, the remaining 53 non-hotspot countries contain only 7812 endemic species. Species-area regression residuals revealed the highest relative index of plant species endangerment in tropical America, tropical Asia, and Southern Africa (Madagascar and South Africa) (Figure 1a). In general, hotspot countries (Figure 1b) have a higher level of current plant species endangerment than non-hotspot ones (Figure 1c). In the tropics, countries in Central and South America, and Southeast Asia generally have a higher level of plant species endangerment than African countries (Figure 1d), while among temperate countries, South Africa, China and Australia were found to have a high level of plant species endangerment (Figure 1e). The five countries with the highest endangerment are (in descending order): Papua New Guinea, New Caledonia, South Africa, Indonesia, and Colombia (Table S2 in Appendix 2). 16 Figure 1. Global map showing the current plant endangerment index of (a) all countries, (b) hotspot countries, (c) non-hotspot countries, (d) tropical countries, and (e) temperate countries. Countries with missing data are unshaded in (a). Map uses a cylindrical equal-area projection. 17 2.3.2. Exposure of threatened plant biodiversity to future habitat loss 2.3.2.1. Habitat loss owing to climate change In 118 hotspot countries where data on future habitat loss and endemism were available, the index of current plant species endangerment was positively correlated with future climate change-driven habitat loss averaged across the four scenarios (Spearman‘s ρ = 0.294, P = 0.001) (Figure 2a). The plant species endangerment index was also positively correlated with climate change-driven habitat loss projected under each of the four separate scenarios (Spearman‘s ρ = 0.232 to 0.330, P = 20th percentile values of plant species richness, historical habitat loss, and future human population growth, covering a total area of 48,097,651 km2 (total area of 756 ecoregions: 130,873,146 km2). Most of these ecoregions are tropical and subtropical moist broadleaf forests (143 ecoregions, 35.9 %), followed by tropical and subtropical dry broadleaf forests (41, 10.3 %), and deserts and xeric shrublands (41, 10.3 %). The majority are situated in the Neotropical (117, 29.4 %) and Indo-Malayan (67, 16.8 %) realms. Trends were similar using the 40th and 60th percentile. One hundred seventyeight ecoregions had > 40th percentile values of plant species richness, historical habitat loss, and future human population growth, covering a total area of 17,059,935 km2. Most of these ecoregions were tropical and subtropical moist broadleaf forests (76: 42.7 %), followed by tropical and subtropical dry broadleaf forests (22: 12.4 %). Most of the 178 ecoregions were found in the Neotropical (63: 35.34 %) and IndoMalayan (36: 20.2 %) realms. For the 60th percentile threshold, the number of ecoregions was reduced to 45 and covered a total area of 5,074,602 km2. Like the ecoregions designated by the 40th percentile threshold, most were tropical and subtropical moist broadleaf forests (19: 42.2 %), followed by tropical and subtropical grasslands, savannas, and shrublands (6: 13.3 %), and tropical and subtropical dry broadleaf forests (5: 11.1 %). The number of ecoregions from the Neotropical (18: 40 %) and Indo-Malayan (10: 44.4 %) tops this group. There were only two high-priority ecoregions when I set the threshold at the 80th percentile. The ecoregion-by-ecoregion findings are presented in Table S5 of Appendix 2. The number of ecoregions and their breakdown by biome and realm type in each 43 category of relative conservation assessment need are presented in Figure S3 and Figure S4 of Appendix 2, respectively. 44 Figure 6. Relative need for the conservation assessment of plant species among 756 global ecoregions (a). Ecoregions are placed in 9 categories ― from 1 = Lowest need (pale yellow) to 9 = Highest need (dark brown). Ecoregions that are ranked above the 80th percentile in plant species richness, historical habitat loss, and future human population increase are placed in category 9. Ecoregions ranked above the 70th percentile but lower than 80th percentile in all three criteria are placed in category 8 and so on. My scheme is overlaid with the BH template (b) and the G200 template (c) for comparison. Map uses a cylindrical equal-area projection. 45 3.3.2. Species-area models used in ranking plant species richness The power model with a log-normal sampling (error) distribution was the top-ranked model in six out of the 13 biomes according to wDIC, but no SAR model was consistently top-ranked across all biomes. The monod log-normal model was topranked in four biomes, followed by the exponential normal model in two biomes, and the negative exponential log-normal model in one biome (Table 4). Models with a log-normal error distribution were top-ranked in 11 out of the 13 biomes. Individual and model-averaged SAR curves are presented in Figure 6. Table 4. DIC weights for species-area relationship models fitted to each biome. The weight of a model can be interpreted as the probability of the model in being the best to fit the data. The weight for the top-ranked model of each biome is highlighted in boldface. ‗DNC‘ refers to models which did not converge in the Markov Chain Monte-Carlo (MCMC) procedure implemented in WinBUGS. SAR Models Error Distribution Biome Normal Power Exp -13 Log-normal Neg Exp -16 1 Neg Exp Monod -20 5.3×10-15 2 2.8×10-10 2.2×10-9 5.7×10-9 5.3×10-9 0.018 0.051 0.931 3 0.050 0.003 0.005 0.011 0.925 0.001 0.004 4 5.6×10 -5 -4 2.0×10 -4 -4 0.905 -3 0.092 1.3×10 -7 5.3×10 -8 -8 0.878 0.021 0.101 5.0×10 -4 0.002 0.956 -4 0.011 2.6×10 -9 -8 0.963 0.002 0.035 6.0×10 -5 -5 0.012 0.191 0.797 -7 2.6×10 6 0.010 0.020 7 3.5×10 -7 -8 2.7×10 -5 8 3.9×10 -5 6.9×10 4.6×10 Power -17 7.7×10 1.4×10 1.0×10 Monod 1 5 3.4×10 -21 3.5×10 9.3×10 1.6×10 5.4×10 9.9×10 2.2×10 5.5×10 0.470 0.524 DNC DNC 0.007 DNC DNC 10 3.7×10-5 7.1×10-5 3.2×10-5 2.9×10-5 0.361 0.250 0.388 11 0.162 0.350 0.140 0.178 0.155 0.005 0.010 12 8.3×10-5 1.8×10-4 9.2×10-4 4.8×10-4 0.004 0.740 0.255 13 7.9×10-15 1.1×10-14 2.5×10-14 2.5×10-14 0.008 0.330 0.662 9 46 Figure 7. Individual and model-averaged special-area relationship (SAR) models fitted using a Markov Chain Monte-Carlo (MCMC) procedure implemented in WinBUGS. SAR models with normal and log-normal sampling distributions were fitted to ecoregions in tropical and subtropical moist broadleaf forests (a), tropical and subtropical dry broadleaf forests (b), tropical, and subtropical coniferous forests (c), temperate broadleaf and mixed forests (d), temperate coniferous forests (e), boreal forests/taiga (f), tropical and subtropical grasslands, savannas, and shrublands (g), temperate grasslands, savannas, and shrublands (h), flooded grasslands and savannas (i), montane grasslands and shrublands (j), tundra (k), Mediterranean forests, woodlands, and scrub or sclerophyll forests (l), and deserts and xeric shrublands (m). The color legend follows Figure 7m. 47 48 49 50 3.3.3. Overlap with existing schemes Among the 398 ecoregions selected using the 20th percentile threshold, 211 (53.0 %) overlap with the ecoregions identified in WWF‘s G200; 292 (73.4 %) are contained in the BH, whilst 75 were excluded from both G200 and BH templates (Table S6 in Appendix 2). Among the 178 ecoregions selected using the 40th percentile threshold, 102 (57.3%) overlap with those designated in the G200, and 148 (83.1%) are contained in BH. Twenty-one ecoregions selected in my analysis were excluded from both G200 and BH (Table 5). Using the 60th percentile threshold, 27 of 45 (60%) ecoregions overlap with the G200 and 42 (93.3%) overlapped with BH. Only two ecoregions were excluded by both G200 and BH — the South Malawi montane forest-grassland mosaic (AT1014) and the Madara Plateau mosaic (AT0710) (Figure S5 in Appendix 3). The two ecoregions selected using the 80th percentile threshold overlap with both BH and the G200. 51 Table 5. List of 21 ecoregions excluded from both Biodiversity Hotspots and Global 200 prioritization template but included in my prioritization template (40th percentile threshold). CODE refers to the WWF ecoregion code of each ecoregion. CAT refers to the prioritization category outlined in Methods. AREA is the area in squarekilometers of each ecoregion. PHBL is the area of natural habitat lost, expressed as a proportion of total terrestrial area of each ecoregion. PDC is human population density change (% change per annum). RPLT is the number of plant species in each ecoregion. CODE CAT AREA PHBL PDC AT0710 AT1014 RPLT 7 7479 0.978 2.181 600 7 10191 0.362 1.938 1900 AA0110 6 1608 0.274 2.130 3200 AT1010 6 13281 0.876 2.454 1300 NA0405 6 52583 0.549 1.230 1553 NA0513 6 3879 0.342 1.286 951 NA0701 6 80515 0.485 1.177 2180 NA0802 6 399039 0.638 1.219 1464 NA0814 6 50215 0.765 1.230 1531 NT0164 6 2071 0.529 1.286 1034 PA0506 6 20071 0.188 2.803 1300 PA1309 6 139333 0.214 3.121 1700 AA0103 5 2810 0.129 2.688 1500 AA0108 5 2407 0.118 2.691 1500 AA0115 5 134636 0.222 2.280 4400 AA0116 5 23184 0.126 2.606 3000 AA0125 5 4180 0.161 2.130 2800 IM0201 5 239409 0.580 0.954 1800 NA0523 5 140797 0.222 0.930 1729 NA0808 5 81929 0.264 2.048 1197 NA0813 5 46993 0.444 1.116 1290 3.3.5. Lack of financial resources in ecoregions important for conservation assessments At 20th to 70th thresholds, the GLM containing per capita wealth was always ranked higher than the null model (Akaike‘s model weights, wAICc ranging from 0.934 to ~1; Table 6). At all thresholds, increasing per capita wealth was negatively correlated with the odds of a country overlapping with ecoregions deemed important for conservation 52 assessments. I did not model the importance status of the 10th and 80th percentile thresholds owing to small sample sizes (four and two countries which did not overlap with important ecoregions, respectively). Country-level results are given in Table S7 in Appendix 2. 53 Table 6. Generalized linear models (GLM) investigating the correlation of per capita wealth with whether a country overlaps with important ecoregions selected using multiple percentile thresholds (a–f). For example, important ecoregions identified using a 20th percentile threshold, are ranked above the 20th percentile in plant species richness, historical habitat loss, and future human population. GLM are ranked by Akaike‘s model weights (wAICc) which represents the probability of the model being the best in the candidate set. ‗gni‘ refers to per capita GNI-PPP. Shown also are the number of parameters (K), log-likelihood (LL), Akaike‘s Information Criterion corrected for small sample size (AICc), difference between the top-ranked models AICc and that of the model under consideration (∆AICc), and percent deviance explained by each model (%DE). a - 20th percentile threshold Rank 1 Structure ~ gni K LL 2 –62.38 2 ~1 (null model) 1 –83.23 AICc ∆AICc wAICc 128.48 0 1 168.48 39.65 %DE 25.05 2.5 × 10-9 b - 30th percentile threshold Rank 1 Structure ~ gni K LL 2 –87.34 2 ~1 1 AICc ∆AICc wAICc 178.76 0 1 –104.73 211.49 32.73 %DE 16.60 7.8 × 10-8 c - 40th percentile threshold Rank 1 Structure ~ gni K LL 2 –98.93 2 ~1 1 AICc ∆AICc wAICc 201.93 0 1 –115.00 232.03 30.10 %DE 13.98 2.9 × 10-7 d - 50th percentile threshold Rank 1 Structure ~ gni K LL AICc ∆AICc wAICc 2 –104.87 213.80 0 1 2 ~1 1 –114.72 231.48 17.68 %DE 8.60 1.5× 10-4 e - 60th percentile threshold Rank 1 Structure ~ gni K LL 2 –97.97 AICc ∆AICc wAICc 200.01 0 0.993 54 %DE 5.75 2 ~1 1 –103.94 209.91 9.91 0.007 f - 70th percentile threshold Rank 1 Structure ~ gni K LL 2 –73.62 5 ~1 1 3.4. –77.28 AICc ∆AICc wAICc 151.30 0 0.934 156.59 5.29 %DE 4.75 0.066 DISCUSSION My study incorporates future human population pressure explicitly into a categorization of the relative need for global-scale conservation assessments, and in this way my approach incorporates both past and future (latent) threat (sensu Cardillo et al. 2006). My approach assumes that species in highly biodiverse ecoregions that have and will continue to experience high habitat degradation are most likely to have the largest numbers of species that are declining in numbers and range, hence those most susceptible to extinction. These ecoregions thus require immediate attention in terms of assessing the population number and the distribution of native plant species to be able to quantify the relative threats to biodiversity. Instead of setting a single and arbitrary threshold, I allocated ecoregions to nine levels of relative need so that conservation assessment managers in the IUCN (and other organizations involved in conservation assessments) can make sound decisions in selecting focal ecoregions for species assessment projects — moving down the percentiles as funds and logistics permit. By modelling the SAR and ranking the species richness in ecoregions within each biome, I ensure a more representative template akin to the G200 (Olson et al. 2001). My template showed moderate spatial congruence with both G200 and Biodiversity Hotspots (Myers et al. 2000), but identifies some important differences. The overlap between the ecoregions 55 selected by my algorithms and the Biodiversity Hotspots template was greater than with the G200 template. A plausible explanation is that the Biodiversity Hotspots template shares my emphasis of conserving plant species and choosing areas with high historical habitat loss. Another reason may be that the Biodiversity Hotspots template covers more ecoregions (451) than the G200 (369). I identified some ecoregions that merit conservation attention that are currently excluded from the list of high-priority areas designated by one or both templates. For example, the South Malawi montane forestgrassland mosaic (WWF eco-code: AT1014) is ranked above the 60th percentile in terms of plant species richness among ecoregions of its biome, historical habitat loss, and future population increase, but it has not been identified by either Biodiversity Hotspots nor the G200 template. The region has had 36 % of its natural habitat already altered, and its human population density is projected to increase by almost 2 % annually to 2015. Countries that overlap with ecoregions of high relative need for conservation assessments are relatively poorer; therefore, funds may not be locally available for the implementation of conservation assessment programs in ecoregions facing high habitat loss. I urge international funding agencies such as the World Bank and Global Environment Facility (GEF) to cooperate with the IUCN and various botanic gardens to facilitate the transfer of funds and scientific expertise from wealthier to poorer countries across high-importance ecoregions in need of conservation assessments. Plant Red List assessments in southern Africa (Siebert & Smith 2005) and other Biodiversity Hotspots (Missouri Botanical Gardens 2009) are examples of projects arising from international cooperation and funding. These multilateral projects also contribute to the long-term 56 conservation of plant species by transferring scientific and management knowledge to local conservation practitioners. I do not explicitly incorporate cost into my prioritization scheme owing to the lack of data on the cost of plant conservation assessments across a range of countries (Siebert & Smith 2005). Moreover, the cost of conservation assessments tend to be flexible depending on the decision to include or exclude non-essential components such as disseminating information at no-charge in the form of books and CDs (Siebert & Smith 2005). However, my highly-important ecoregions — like the high-priority watersheds identified using an ecosystem-services approach (Luck et al. 2009) to prioritization — are located in areas characterized by a high conservation benefit-to-cost ratio (i.e., developing countries with low per capita wealth; Balmford et al. 2003). This observation argues for even greater motivation for investment in these areas. Cooperation among international funding agencies, botanic gardens, and local scientists is imperative for the implementation of conservation assessment programs in highly biodiverse and highly threatened ecoregions. Lastly, I considered multiple SAR models in a Bayesian framework to identify ecoregions of high species richness, therefore accounting explicitly for uncertainty in the underlying distributions. Species richness across ecoregions is log-normally distributed in most biomes, whereas all previous studies have assumed a normal distribution (e.g., Fattorini 2006; Guilhaumon et al. 2008). My results highlight the importance of considering a wide variety of SAR models with different error distributions to identify species-rich hotspots, and I provided a possible approach using Bayesian statistics. 57 4. CONCLUSIONS The central question I aimed to answer in this thesis was: What conservation strategies can we employ to prevent or slow down the loss of plant species? To answer this question, I first elucidated the distribution of future habitat loss with respect to the current distribution of threatened species at the country level in Chapter 2. Habitat loss driven by climate change is likely to be highest in countries with the highest level of plant species endangerment. The association between climate change impacts and threatened plant biodiversity demands for greater political action in reducing greenhouse gas emissions. In particular, the inclusion of a REDD mechanism in the UNFCCC may reduce global carbon emissions, and allow developing countries with high conservation need to improve their economic condition, as well as, mitigate habitat and species loss. By considering the distribution of currently threatened species and future habitat loss, countries located near the equator are expected to have highest level of plant species endangerment in the future. Although the current conservation assessment effort is correlated with future plant species endangerment, I urge the acceleration of conservation assessments in countries where the level of future plant species endangerment is projected to be high because only ~5% of all plant species have been assessed. Poverty and poor governance may exacerbate the impact of high habitat loss on threatened plant biodiversity. I incorporated economic condition and governance quality with the level of future plant species endangerment to assess the relative conservation need among countries. The Democratic Republic of Congo, Ethiopia, Nepal, and Tajikistan were the countries found to have the greatest conservation need owing to their low quality of governance, poor economic condition, and high relative level of future plant species 58 endangerment. Politicians and conservation managers can therefore formulate sound conservation policies such as disbursing financial aid to economically-challenged countries or improving local governance and administrative capacity in governmentallyweak countries that are at the same time prone to high levels of future species loss. In the local context, novel approaches that aim to improve economic condition and environmental governance, and at the same time, yield ecological benefits like the deliberative community-based conservation approach should be implemented in the future. After discussing a range of possible global and local strategies to conserve plant biodiversity in the face of projected climate change and land-use change, I focused on developing one specific strategy in Chapter 3 — identifying the most important ecoregions for conservation assessments. Identifying important areas for conservation has become a central tenet of conservation biology because resources available for conservation are limited. Conservation assessment programs are relatively inexpensive and provide important baseline data on the range and distribution of species. These programs also provide a yardstick against which success of existing and future conservation efforts can be measured. I identified speciose and highly-threatened ecoregions to be prioritized for conservation assessments. We also highlighted the importance of several ecoregions that were left out by existing conservation prioritization schemes (e.g., the South Malawi montane forest-grassland mosaic [AT1014] and the Madara Plateau mosaic [AT0710]). As high priority areas are located in countries that are less wealthy, I urge international aid agencies and botanic gardens to cooperate with local scientists to fund and implement conservation assessment programs. In providing a new 59 statistical framework to incorporate the uncertainty of the SAR models and error distribution in ranking species richness of ecoregions, we found that species richness was log-normally distributed across most biomes but there was no single best SAR model across all biomes. My results highlighted the importance of considering a wide variety of SAR models with different error distributions to identify species-rich hotspots. 60 REFERENCES Alcamo, J., G. J. J. Kreileman, M. Krol, R. Leemans, J. Bollen, J. van Minnen, M. Shaefer, S. Toet, B. de Vries. 1998. Global modeling of environmental change: an overview of IMAGE 2.1. in Alcamo, J., R. Leemans, G. J. J. Kreileman (eds.) Global change scenarios of the 21st century. Pergamon/Elsevier, Oxford. Anderson, D. R. 2008. Model based inference in the life sciences: a primer on evidence. Springer, New York. Arrhenius, O. 1921. Species and area. Journal of Ecology 9: 95–99. Balmford, A., A. Long. 1995. Across country analyses of biodiversity congruence and current conservation effort in the tropics. Conservation Biology 9: 1539–1547. Balmford, A., K. J. Gaston, S. Blyth, A. James, V. Kapos. 2003. Global variation in terrestrial conservation costs, conservation benefits, and unmet conservation needs. Proceedings of the National Academy of Sciences of the United States of America 100: 1046–1050. Bolker, B. M. 2008. Ecological models and data in R. Princeton University Press, Princeton. Bojö, J., R. Chandra Reddy. 2001. Poverty reduction strategies and environment: A review of 40 interim and full PRSPs. World Bank Environment Department, Geneva. Bradshaw, C. J. A., X. Giam, H. T. W. Tan, B. W. Brook, N. S. Sodhi. 2008. Threat or invasive status in legumes is related to opposite extremes of the same ecological and life history attributes. Journal of Ecology 96: 869–883. 61 Bradshaw, C. J. A., N. S. Sodhi, B. W. Brook. 2009a. Tropical turmoil: a biodiversity tragedy in progress. Frontiers in Ecology and the Environment 7: 79–87. Bradshaw, C.J.A., I. G. Warkentin, N. S. Sodhi. 2009b. Urgent preservation of boreal carbon stocks and biodiversity. Trends in Ecology and Evolution. doi: 10.1016/j.tree.2009.03.019. Bramwell, D. 2002. How many plant species are there? Plant Talk 28: 32–34. Brook, B. W., N. S. Sodhi, P. K. L. Ng. 2003. Catastrophic extinctions follow deforestation in Singapore. Nature 424: 420–423. Brooks, T. M., R. A. Mittermeier, G. A. B. Da Fonseca, J. Gerlach, M. Hoffmann, J. F. Lamoreux, C. G. Mittermeier, J. D. Pilgrim, A. S. L. Rodrigues. 2006. Global biodiversity conservation priorities. Science 313: 58–61. Bruner, A. G., R. E. Gullison, A. Balmford. 2004. Financial costs and shortfalls of managing and expanding protected-area systems in developing countries. Bioscience 54: 1119–1126. Burnham, K. P., D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach ― 2nd edition. Springer, New York. Cam, E., J-Y. Monat, J. A. Royle. 2004. Dispersal and individual quality in a long lived species. Oikos 106: 386-398. Cardillo, M., G. M. Mace, J. L. Gittleman, A. Purvis. 2006. Latent extinction risk and the future battlegrounds of mammal conservation. Proceedings of the National Academy of Sciences of the United States of America 103: 4157-4161. CIESIN (Center for International Earth Science Information Network) & CIAT (Centro Internacional de Agricultura Tropical). (2005). Gridded Population of the World 62 Version 3 (GPWv3) [WWW document]. URL http://sedac.ciesin.columbia.edu/gpw. Accessed 10 December 2008. Daily, G. C. 1997. Nature's services. Washington, D.C.: Island Press. Davies, R. G., C. D. L. Orme, V. Olson, G. H. Thomas, S. G. Ross, T-S. Ding, P. C. Rasmussen, A. J. Stattersfield, P. M. Bennett, T. M. Blackburn, I. P. F. Owens, K. J. Gaston. 2006. Human impacts and the global distribution of extinction risk. Proceedings of the Royal Society B: Biological Sciences 273: 2127–2133. Dechezleprêtre, A., M. Glachant, Y. Meniérè. 2007. The North-South transfer of climatefriendly technologies through the Clean Development Mechanism. CERNA, École des Mines de Paris. Deutz, A. M. 2005. Biodiversity: Biodiversity governance after Johannesburg. Pages 185–202 in G. Ayre, R. Callway, editors. Governance for sustainable development: A foundation for the future. Earthscan, London, United Kingdom. Ehrlich, P. R., R. M. Pringle. 2008. Where does biodiversity go from here? A grim business-as-usual forecast and a hopeful portfolio of partial solutions. Proceedings of the National Academy of Sciences of the United States of America 105: 1157911586. Fattorini, S. 2006. Detecting biodiversity hotspots by species-area relationships: A case study of Mediterranean beetles. Conservation Biology 20:1169-1180. Fattorini, S. 2007. To fit or not to fit? A poorly fitting procedure produces inconsistent results when species-area relationship is used to locate hotspots. Biodiversity and Conservation 16: 2531–2538. 63 Franklin, A. B., D. R. Anderson, K. P. Burnham. 2002. Estimation of long-term trends and variation in avian survival probabilities using random effects models. Journal of Applied Statistics 29: 267–287. Gaveau, D. L. A., M. Linkie, Suyadi, P. Levang, N. Leader-Williams. 2009. Three decades of deforestation in southwest Sumatra: Effects of coffee prices, law enforcement and rural poverty. Biological Conservation 142: 597–605. Geist, H. J., E. F. Lambin. 2002. Proximate causes and underlying driving forces of tropical deforestation. Bioscience 52: 143–150. Govaerts, R. 2001. How many species of seed plants are there? Taxon 50: 1085. Guilhaumon, F., O. Gimenez, K. J. Gaston, D. Mouillot. 2008. Taxnomic and regional uncertainty in species-area relationships and the identification of species richness hotspots. Proceedings of the National Academy of Sciences of the United States of America 105: 15458-15463. Hamilton, A., P. Hamilton. 2006. Plant conservation: an ecosystem approach. Earthscan, London. Hansen, M. C., R. S. DeFries. 2004. Detecting long-term global forest change using continuous fields of tree-cover maps from 8-km Advanced Very High Resolution Radiometer (AVHRR) data for the years 1982-1999. Ecosystems 7: 695-716. Hobohm, C. 2003. Characterization and ranking of biodiversity hotspots: centres of species richness and endemism. Biodiversity and Conservation 12: 279-287. Hoekstra, J. M., T. M. Boucher, T. H. Ricketts, C. Roberts. 2005. Confronting a biome crisis: global disparities of habitat loss and protection. Ecology Letters 8: 23–29. 64 Huston, M.A. 1994. Biological Diversity: The coexistence of species on changing landscapes. Cambridge University Press, Cambridge, UK. Image–Team. 2001. The IMAGE 2.2 implementation of the SRES scenarios: a comprehensive analysis of emissions, climate change and impacts in the 21st century. RIVM (Rijksinstituut voor Volksgezondheid en Milieu/National Institude of Public Health and Environment), Bilthoven, The Netherlands. IPCC (Intergovernmental Panel on Climate Change). 2007. Climate change 2007: impacts, adaptation and vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change (IPCC). Cambridge University Press, Cambridge, UK. IUCN (International Union for Conservation of Nature and Natural Resources). 2008. 2008 IUCN Red List of threatened species. [WWW document]. URL http://www.iucnredlist.org. Accessed March 2009. James, A. N., K. J. Gaston, A. Balmford. 1999. Balancing the Earth's accounts. Nature 401: 323–324. Jetz, W., D. S. Wilcove, A. P. Dobson. 2007. Projected impacts of climate and land-use change on the global diversity of birds. PLoS Biology 5: e157. Jha, S., K. S., Bawa. 2006. Population growth, human development, and deforestation in biodiversity hotspots. Conservation Biology 20: 906–912. Kaufmann, D., A. Kraay, M. Mastruzzi. 2008. Governance Matters VII: Governance indicators for 1996–2007. World Bank Policy Research June 2008. The World Bank Group. 65 Kerr, J. T., D. J. Currie. 1995. Effects of human activity on global extinction risk. Conservation Biology 9: 1528–1538. Kier, G., J. Mutke, E. Dinerstein, T. H. Ricketts, W. Kuper, H. Kreft, W. Barthlott. 2005. Global patterns of plant diversity and floristic knowledge. Journal of Biogeography 32: 1107–1116. Lamoreux, J. F., J. C. Morrison, T. H. Ricketts, D. M. Olson, E. Dinerstein, M. W. McKnight, H. H. Shugart. 2006. Global tests of biodiversity concordance and the importance of endemism. Nature 440: 212-214. Lee, T. M., W. Jetz. 2008. Future battlegrounds for conservation under global change. Proceedings of the Royal Society B: Biological Sciences 275: 1261–1270. Limpert, E., W. A. Stahel, M. Abbt. 2006. Log-normal distribution across the sciences: keys and clues. Bioscience 51: 343-352. Linder, H. P. 2001. Plant Diversity and Endemism in Sub-Saharan Tropical Africa. Journal of Biogeography 28: 169-182. Luck, G. W., K. M. A. Chan, J. P. Fay. 2009. Protecting ecosystem services and biodiversity in the world's watersheds. Conservation Letters., in press. Lunn, D. J., A. Thomas, N. Best, D. Spiegelhalter. 2000. WinBUGS - a Bayesian modelling framework: concepts, structure, and extensibility. Statistics and Compuing 10: 325-337. Mau, B., M. A. Newton, B. Larget. 1999. Bayesian phylogenetic inference via Markov chain Monte Carlo methods. Biometrics 55: 1-12. 66 Miles, K. 2005. Innovative financing: filling in the gaps on the road to sustainable environmental funding. Review of European Community & International Environmental Law 14: 202–211. Millennium Ecosystem Assessment. 2005a. Ecosystems and human well-being: biodiversity synthesis. Island Press. Millennium Ecosystem Assessment. 2005b. Ecosystems and human well-being: scenarios. Island Press. Missouri Botanical Gardens. 2009. Species-level conservation. [WWW document]. URL http://www.mobot.org/plantscience/ccsd/programs/specieslcon.shtml. Accessed 28 July 2009. Mittermeier, R. A., P. R. Gil, M. Hoffmann, J. D. Pilgrim, T. M. Brooks, C. G. Mittermeier, J. F. Lamoreux, G. A. B. Da Fonseca. 2004. Hotspots Revisited. CEMEX. Myers, N., R. A. Mittermeier, C. G. Mittermeier, G. A. B. Da Fonseca, J. Kent. 2000. Biodiversity hotspots for conservation priorities. Nature 403: 853–858. Neumann, R. P. 2001. Disciplining peasants in Tanzania: from state violence to selfsurveillance in wildlife conservation. Pages 305–327 in N. L. Peluso, M. Watts, editors, Violent Environments. Cornell University Press, Ithaca. O'Connor, C., M. Marvier, P. Kareiva. 2003. Biological vs. social, economic and political priority-setting in conservation. Ecology Letters 6: 706–711. Olson, D. M., E. Dinerstein, E. D. Wikramanayake, N. D. Burgess, G. V. N. Powell, E. C. Underwood, J. A. D'amico, I. Itoua, H. E. Strand, J. C. Morrison, C. J. Loucks, T. F. Allnutt, T. H. Ricketts, Y. Kura, J. F. Lamoreux, W. W. Wettengel, P. 67 Hedao, K. R. Kassem. 2001. Terrestrial ecoregions of the world: a new map of life on Earth. BioScience 51: 933–938. Pattyn, F., W. van Huele. 1998. Power law or power flaw? Earth Surface Processes and Landforms 23: 761-767. Pimm, S., P. Raven, A. Peterson, C. H. Sekercioglu, P. R. Ehrlich. 2006. Human impacts on the rates of recent, present, and future bird extinctions. Proceedings of the National Academy of Sciences of the United States of America 103: 10941–10946. Pitman, N. C. A., P. M. Jorgensen. 2002. Estimating the size of the World's threatened flora. Science 298: 989. Prance, G. T., H. Beentje, J. Dransfield, R. Johns. 2000. The tropical flora remains undercollected. Annals of the Missouri Botanical Garden 87: 67–71. Primack, R.B., R. T. Corlett. 2005. Tropical Rain Forests: An Ecological and Biogeographical Comparison. Blackwell Publishing, Oxford, UK. R Development Core Team. 2008. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna. Rao, C. R. 1973. Linear statistical inference and its applications. Wiley, New York. Rao, M., M. Rabinowitz, S. T. Khaing. 2002. Status review of the protected-area system in Myanmar, with recommendations for conservation planning. Conservation Biology 16: 360–368. Ratkowsky, D. A. 1990. Handbook of nonlinear regression models. Marcel Dekker, New York. Scotland, R. W., A. H. Wortley. 2003. How many species of seed plants are there? Taxon 52: 101–104. 68 Siebert, S. J., G. F. Smith. 2005. Plant red data list assessments in Southern Africa: Financial costs of a collaborative regional project. Taxon 54: 1051-1055. Smith, J., K. Obidzinski, Subarudi, I. Suramenggala. 2003. Illegal logging, collusive corruption and fragmented governments in Kalimantan, Indonesia. International Forestry Review 5: 293–302. Sodhi, N. S., B. W. Brook, C. J. A. Bradshaw. 2007. Tropical conservation biology. Blackwell Publishing, Singapore. Sodhi, N. S., D. Bickford, A. C. Diesmos, T. M. Lee, L. P. Koh, B. W. Brook, C. H. Sekercioglu, C. J. A. Bradshaw. 2008. Measuring the meltdown: drivers of global amphibian extinction and decline. PLoS ONE 3: e1636. Sodhi, N.S., Brook, B.W. & Bradshaw, C.J.A. (2007). Tropical conservation biology. Blackwell Publishing, Singapore. Spiegelhalter D. J., N. G. Best, B. P. Carlin, A. Van der Linde. 2002. Bayesian measures of model complexity and fit. Journal of the Royal Statistical Society Series B (Statistical Methodology) 64: 583-616. Stern, P. C. 2005. Deliberative methods for understanding environmental systems. BioScience 55: 976–982. Thomas, C. D., A. M. A. Franco, J. K. Hill. 2006. Range retractions and extinction in the face of climate warming. Trends in Ecology & Evolution 21: 415-416. Thuiller, W., S. Lavorel, M. B. Araujo, M. T. Sykes, I. C. Prentice. 2005. Climate change threats to plant diversity in Europe. Proceedings of the National Academy of Sciences of the United States of America 102: 8245–8250. 69 Underwood, E. C., M. R. Shaw, K. A. Wilson, P. Kareiva, K. R. Klausmeyer, M. F. McBride, M. Bode, S. A. Morrison, J. M. Hoekstra, H. P. Possingham. 2008. Protecting biodiversity when money matters: maximizing return on investment. PLoS ONE 3: e1515. Doi: 10.1371/journal.pone.0001515 Van Vuuren, D. P., O. E. Sala, H. M. Pereira. 2006. The future of vascular plant diversity under four global scenarios. Ecology and Society 11: 25. Wilson, K. J., E. C. Underwood, S. A. Morrison, K. R. Klausmeyer, W. M. Murdoch, B. Reyers, G. Wardell-Johnson, P. A. Marquet, P. W. Rundel, M. F. McBride, R. L. Pressey, M. Bode, J. M. Hoekstra, S. Andelman, M. Looker, C. Rondinini, P. Kareiva, M. R. Shaw, H. P. Possingham. 2007. Conservation biodiversity efficiently: what to do, where, and when? PLoS Biology 5: e223. Doi: 10.1371/journal.pbio.0050223. Wilson, R. J., D. Gutirrez, J. Gutirrez, D. Martinez, R. Agudo, V. J. Monserrat. 2005. Changes to the elevational limits and extent of species ranges associated with climate change. Ecology Letters 8: 113-1146. World Resources Institute. 2007. EarthTrends: Environmental Information. [WWW document]. URL http://earthtrends.wri.org. Accessed February 2009. Wright, S. J. 1981. Intra-archipelago vertebrate distributions: the slope of the species-area relation. American Naturalist 118: 726-748. 70 [...]... and high future plant species endangerment were identified as having the highest conservation need I provided the first global-scale assessment of the association between threatened plant species and future land use- and climate change- driven habitat loss, and present plausible policies towards plant species conservation By considering these projected impacts together with governance quality and poverty,... data from Lee and Jetz (2008) who calculated the percentage of area subjected to land- cover transformation in 174 countries owing to land- use and climate changes over the four MEA socioeconomic scenarios from 2000 to 2050 Lee and Jetz (2008) classified transformation from natural- to humaninduced land cover types as land- use- driven (e.g., pristine forest converted to agricultural land) , and change from... the current level of plant species endangerment to total future habitat loss, I ranked 163 countries (with data available for future habitat loss and current endangerment index) according to their future level of plant species endangerment (Lee & Jetz 2008) Each country was separately ranked in terms of total future habitat loss (from both land- use and climate changes) and current species endangerment... conducted additional analyses using land- cover transformation data from each of the four scenarios to examine the sensitivity of my results towards the dominance of any one particular scenario I used Spearman‘s ρ rank-order correlation to test the correlation between the index of species endangerment and future habitat loss owing to land- use and climate changes 2.2.4 Future plant species endangerment rankings... the level of plant species endangerment and climate change induced habitat loss, as well as human land- use change induced habitat loss, separately Next, I related the current species endangerment level to future habitat loss to determine the future level of endangerment following the ranking procedures of Lee & Jetz (2008) To evaluate the efficacy of the current prioritization pattern of IUCN species. .. important for conservation assessments are poorer compared to others In this thesis, I provided the first global-scale assessment of the association between threatened plant species and future land use- and climate change- driven habitat loss My study is the first attempt to prioritize the conservation assessment of plant species by taking into account uncertainty in the form and distribution of SAR models,... and the metric for future habitat loss owing to land- use change was the average area projected to be converted via human land- use change divided under the four scenarios to year 2050 expressed as a percentage of total land area I used the percentage land- cover transformations averaged over all four scenarios for my main analyses because the ‗actual‘ future is likely to fall in between the four scenarios... under future scenarios, I tested the association between the proportion of species assessed in a country and future plant species endangerment Lastly, I incorporated economic condition and quality of governance with future plant species endangerment ranks to determine countries of high conservation need By considering these projected impacts together with governance quality and poverty, my system of. .. question, what conservation strategies can we employ to prevent or slow 1 down the loss of plant species? This thesis, consisting of two main papers (Chapters 2 and 3), aims to answer this main question as well as other related questions In Chapter 2 (Future Habitat Loss and Threatened Plant Biodiversity1), I assessed the exposure of threatened plant biodiversity to land- use and climate change- driven... natural land- cover category (e.g., pristine forest converted to savanna) as driven by climate change Thus, my metric for climate change- driven habitat 11 loss was the average area projected to undergo transformation from one natural landcover category to another under the four socioeconomic scenarios to the year 2050 expressed as a percentage of the total land area of a country, and the metric for future ... the index of species endangerment and future habitat loss owing to land- use and climate changes 2.2.4 Future plant species endangerment rankings By relating the current level of plant species endangerment... globally threatened plant species 2.2.2 National index of relative plant species endangerment 10 2.2.3 Future habitat loss from land- use and climate changes 11 2.2.4 Future plant species endangerment... owing to land- use change was the average area projected to be converted via human land- use change divided under the four scenarios to year 2050 expressed as a percentage of total land area I used