Using the ancient past for establishing current threat in poorly inventoried regions

This article appeared in a journal published by Elsevier. The attachedcopy is furnished to the author for internal non-commercial researchand education use, including for instruction at the authors institution

and sharing with colleagues.

Other uses, including reproduction and distribution, or selling orlicensing copies, or posting to personal, institutional or third party

websites are prohibited.

In most cases authors are permitted to post their version of thearticle (e.g. in Word or Tex form) to their personal website orinstitutional repository. Authors requiring further information

regarding Elsevier’s archiving and manuscript policies areencouraged to visit:

http://www.elsevier.com/copyright

http://www.elsevier.com/copyright

Author's personal copy

Using the ancient past for establishing current threat in poorly inventoried regions

Craig M. Costion a,b,!, Jolie Liston c, Ann H. Kitalong d, Akiko Iida e, Andrew J. Lowe a,f

a Australian Centre for Ecology and Evolutionary Biology, University of Adelaide, Adelaide SA5005, Australiab Australian Tropical Herbarium, E2 Bld. JCU Cairns Capmus PO Box 6811, Cairns QLD4870, Australiac Department of Archaeology & Natural History, College of Asia and the Pacific, Australian National University, Australiad The Environment Inc. PO Box 1696, Koror 96940, Palaue Department of Urban Engineering, University of Tokyo, Japanf State Herbarium of South Australia, Department of Environment and Natural Resources, Adelaide SA5005, Australia

a r t i c l e i n f o

Article history:Received 22 September 2011Received in revised form 12 December 2011Accepted 20 December 2011Available online 13 January 2012

Keywords:Conservation paleobiologyThreatened speciesIUCNSavannaPalauMicronesia

a b s t r a c t

The need for a global priority list for threatened plants has been widely recognized by the conservationcommunity, yet the threatened status of the majority of the world’s plants species remains poorly known.This is especially true in the tropics and the oceanic islands of the Pacific, where progress towards thetargets of the Global Strategy for Plant Conservation (GSPC) 2011–2020 is hindered by the paucity ofcomplete species distribution data. Here we outline a new methodology to undertake threatened speciesassessments where detailed contemporary population data is lacking. This new interdisciplinary method-ology draws upon the synthesis of archaeological and botanical data to calculate a percentage of long-term decline in habitat quality. We use this method to assess the threatened status of the endemic floraof Palau, Micronesia, a Pacific island nation known for its high levels of plant diversity and endemism, byutilizing data extending back to human colonization of the archipelago. For Palau, we calculate the per-centage of a long-term decline in habitat quality to be 31–39% of the total available range of 55% of theendemic plant species. These data are also used to address a long debated question in the western Pacific:Are the origins of the savanna vegetation anthropogenic? Strong evidence for anthropogenic savannas inMicronesia support the estimated extent of historic deforestation in Palau. This new method worked wellin our case study, and can be used in other locations with incomplete species distribution data toestablish a first basis for conservation prioritization.

! 2011 Elsevier Ltd. All rights reserved.

1. Introduction

In 2002 the Convention on Biological Diversity (CBD) adopted aGlobal Strategy for Plant Conservation (GSPC), which targeted apreliminary assessment of the conservation status of all knownplant species by 2010 (Baillie et al., 2004). To date in Oceania,excluding Australia, New Zealand and Hawai’i, there have been atotal of 1304 IUCN red list assessments completed (IUCN, 2010)which have been estimated to comprise approximately 9% of thetotal known and described plant diversity in the Pacific islandsregion (Pippard, 2008). Up to date data on the total number ofindigenous plant species from the major Pacific islands and archi-pelago regions suggests this estimate is over-estimated by approx-imately 6% (Table 1). Of all the regions where accurate data isavailable, only 3% out of the total number of species have beenassessed with IUCN Criteria to date suggesting that approximately

97% of the plant species native to underdeveloped countries of thePacific Island bioregion await IUCN threat assessment. These per-centages are consistent with the total percent of the world’s plantspecies assessed, estimated at 3–4% (Jennings and Rohr, 2011; Riv-ers et al., 2011).

The CBD reconvened in 2010 to revise the original targets andextended the GSPC targets to 2020. New targets include aims for‘‘an assessment of the conservation status of all known plant spe-cies, as far as possible, to guide conservation action’’ (Secretariat ofthe Convention on Biological Diversity, 2011). Although the moredeveloped nations of Oceania, including Australia, New Zealand,and the US state of Hawai’i all have independent legislation forassigning and monitoring threatened status of species, the IUCNCriteria are still widely promoted as both a global standard forassessing threat and for application in regions where national leg-islation is lacking. In this study we argue strongly that in order toeven approach the ambitious GSPC 2020 target, alternative meth-ods of assessing threatened status of plant species are absolutelynecessary. Here we present one such alternative method which isspecifically designed for application in poorly inventoried florassuch as Oceania. Our method is outlined in seven steps using a casestudy from Palau, Micronesia.

0006-3207/$ - see front matter ! 2011 Elsevier Ltd. All rights reserved.doi:10.1016/j.biocon.2011.12.026

! Corresponding author at: Australian Centre for Ecology and EvolutionaryBiology, University of Adelaide, Adelaide SA5005, Australia. Tel.: +61 (0)7 40421837; fax: +61 (0)7 4042 1842.

E-mail address: [email protected] (C.M. Costion).

Biological Conservation 147 (2012) 153–162

Contents lists available at SciVerse ScienceDirect

Biological Conservation

journal homepage: www.elsevier .com/ locate /biocon


Considering the vulnerability of islands to biological extinc-tions, the lack of red list assessments to date in the Pacific is of con-cern. Of the world’s recorded extinctions for mammals, birds,amphibians, reptiles and mollusks, 72% were island species (Baillieet al., 2004) suggesting that the Polynesia-Micronesia BiodiversityHotspot (Myers et al., 2000), though not as rich in species as manyother global biodiversity hotspots, is potentially more vulnerableto extinctions. It is probable that the loss of many species of Ocea-nia’s flora and fauna have gone unrecognized and unrecorded.Many studies have indicated that entire assemblages of species be-came extinct following the colonization of the Pacific by humans.Extinctions associated with early settlement of the Pacific havebeen well documented in the palaeorecord for plants (Prebbleand Dowe, 2007), birds (Holdaway, 1989; Athens et al., 2002;Steadman, 2006), amphibians (Holdaway, 1989), reptiles (Pregilland Steadman, 2004) and insects (Porch, 2007; Porch et al., 2007).

Despite this wealth of palaeoenvironmental and archaeologicaldata, current IUCN Criteria only enable assessments to draw fromevidence extending ten years or three generations into the past.Here we argue that in the Pacific, and likely other poorly knowntropical regions, the palaeoenvironmental and archaeological re-cord is often the best evidence available for identifying populationdeclines of poorly known species. Failure to utilize and draw infer-ences from these disciplines to inform conservation policy willonly ensure the continued slow progress towards threat assess-ment and in some cases may come too late to help save speciesfrom extinction. To test out this new methodology, the Pacific is-land nation of Palau is used as a case study, where evidence for

an anthropogenic origin of a contended vegetation type revealedmeasurable declines in the populations of its endemic plantspecies.

1.1. Background – Palau overview

Palau is an archipelago of more than 350 islands extending150 km along a north to southwest-trending arc in the westernCaroline Islands approximately 900 km north of West Papua and870 km east of the Philippines at the western limit of the Polyne-sia–Micronesia Biodiversity Hotspot. Palau has tropical broadleaved evergreen moist forest occurring on several larger volcanicislands, and numerous atoll, platform-like, and uplifted karst lime-stone islands. The archipelago is regarded for having the most di-verse flora of all the Micronesian islands in number of bothnative and endemic plant species. This study focuses on the 75plant species restricted to Palau’s 375.89 km2 of volcanic islands(Fig. 1) and draws upon evidence from Palau’s palaeoenvironmen-tal, archaeological and historical records to quantify an inferredpast decline in native forest habitat.

A preliminary IUCN red list assessment of Palau’s endemicplants was conducted (Costion et al., 2009), however detailed dis-tribution data was lacking for 61% of the species, and were catego-rized as ‘‘DD’’ (Data Deficient). Here we re-assesses the Palauendemic flora with our new proposed methodology that enablesa complete assessment with the existing available data.

Substantial deforestation is identified in Palau’s palaeorecord byan increase in savanna indicators along with a rise in the frequency

Table 1Percent of the native flora assessed under IUCN Criteria for respective regions of Oceania (See Supplementary Data 1 for complete list of data sources).

Region Native plants (vascular) % Endemism of native flora IUCN plant assessments % of Native flora assessed

New Guinea 25 000+ !90% 317 1Norfolk Island 136 32% 3 2Samoa 550 30% 20 4Tonga 340 4% 16 5Solomon Islands 1941 Unknown 88 5Vanuatu 870 !15% 46 5Micronesia 1227 30% 97 8Fiji 1594 54% 141 9New Caledonia 3137 79% 346 11Total 34 795 – 1074 3

Fig. 1. Geographical range and total area of occupancy of the taxa under assessment.

154 C.M. Costion et al. / Biological Conservation 147 (2012) 153–162


of microscopic charcoal particles between about 4400/3000 and3800/2500 cal yr BP (Athens and Ward, 2002, 2005; Athens andStevenson, 2010) and in the form of remnant terrace formations(Fig. 2) derived from earthwork construction and use that beganover 2000 years ago (Liston, 2007, 2009). During Palau’s Japaneseera (1914–1944), Babeldaob’s land-cover also experienced signifi-cant and widespread human-induced impact (Iida, 2011) fromthe production of cash crops, harvesting of timber, and strip-min-ing. These anthropogenic activities which would have stripped theforest bare, provide dated evidence for forest succession studies(Liston et al., 2011). Although the cause of transition from foresthabitat to savanna in Micronesia has remained a hotly debated to-pic, here we provide new evidence to support an anthropogenicorigin of Palau’s savanna. Our demonstrated utilization of this evi-dence of long term decline in habitat quality for threatened speciesassessments has substantial implications for progressing biodiver-sity conservation at local, regional and global scales.

2. Methods

IUCN Criterion A is based on reductions of species populationsthat can be observed, estimated, inferred or suspected in the pastor future; however, these population declines must be measuredin periods of ten years or three generations (IUCN, 2001). In an at-tempt to address the serious data deficiency gaps in the IUCN redlist, we remove the limitation imposed by the 10 year/3 generationmeasurable period of decline. This enables archaeological and pal-aeobotanical evidence to be incorporated into a threatened speciesassessment. These data have the potential to identify historicaldeclines and/or fluctuations in habitat quality coinciding withhuman activities and can directly indicate measurable changes inextant populations of endemic species. Our methodology has thefollowing seven step protocol (Fig. 3).

(1) Quantify area of occupancy based on plant species endemicstatus and habitat restrictions (Costion et al., 2009). Loss orfluctuation of habitat within this area can then be directlyattributed to population declines of the endemic species.

All plants classified as endemic to volcanic substrates areconsidered. Following IUCN Criterion B2 we define three cat-egories of area occupancy: (1) <10 km2, (2) <500 km2 and (3)<2000 km2. Taxa placed in each category can then beweighted or prioritized in numerical order relative to theirtotal population size. This enables prioritization to beassigned to taxa with extremely small natural ranges orareas of occupancy. All taxa in the present study arerestricted to an area of occupancy of 375.89 km2 (Fig. 1) thusreceive priority level (2).

(2) Conduct a thorough review of the islands’ archaeological andbotanical literature including documentation of the pollenrecord, archaeological investigations and detailed speciescomposition of vegetation types. This allows for determina-tion of the likely extent of forest habitat on the archipelagoprior to colonization and established evidence of landscapechanges that coincided with the initial stages of human set-tlement (see Athens and Ward, 2005).

(3) Formulate or build upon an existing or contended hypothe-sis on the anthropogenic origins of vegetation type/s by syn-thesizing data from multiple disciplines. In the case ofMicronesia, savanna.

(4) Test the hypothesis to confirm the anthropogenic origin ofspecific vegetation type/s. For Palau this involved: (1)archaeological and historical evidence of deforestation (SeeSection 3.3.1), which included a review of literature and car-tographic documents, fieldwork and remote sensing meth-ods relying on historic vegetation maps and aerialphotography (see Supplementary Data 2), and; (2) taxo-nomic evidence for the antiquity of the vegetation type.We define taxonomic evidence (See Section 3.3.2) as theabsence of valid endemic plant species restricted to savannavegetation. The endemic status of plants allegedly endemicto the savannas of Micronesia were assessed by examiningtheir distribution records and relevant taxonomic literature.

(5) Quantify the extent of past identifiable habitat decline andrelate it to current land cover by (1) calculating the area ofdemonstrable anthropogenic disturbance and; (2) overlay-ing the data onto existing vegetation maps. To accuratelyreflect the complexity of Babeldaob’s landscape history, wereviewed documented evidence of forest re-growth andrecovery in areas where forest habitat loss was directly ver-ifiable. The extent of current intact and disturbed volcanicisland habitats was classified (see Supplementary Data 6)using data from Palau’s 1976 aerial photograph-based vege-tation survey (Cole et al., 1987).

(6) Total inferred decline in habitat is used to qualify speciesrestricted to the habitat of assessment as threatened wherea P30% decline in habitat = At Risk, P50% = Threatened withExtinction, and P80% = Extremely Threatened. We furtherdefine these categories as follows:At Risk: Species are at risk of extinction and require legisla-tive protection of representative populations and regularmonitoring at appropriate intervals to report on theirstability.Threatened: Species are under a threat of extinction signifi-cant enough to warrant conservation actions and plans tomitigate and revert the ongoing decline.Extremely Threatened: Species require immediate actions andintervention to avert extinction threat.

(7) Finally, we discuss how landscape or land-use history caninfluence interpretations of current vegetation types andthe implications that this has for species-based conservationpolicy. We propose that species from hotspot or tropicalregions assessed using the above methodology qualify forglobally recognized threatened status.

Fig. 2. Aerial photograph of part of Ngaremlengui Earthwork District (USGS aerialphoto 4_46_2_244 – 1976).

C.M. Costion et al. / Biological Conservation 147 (2012) 153–162 155


3. Results and discussion

Our results are presented following the seven-stepmethodology.

3.1. Establishing endemic status and habitat restrictions

The most important step in the assessment process, and a highpriority in the Pacific, is an up-to-date checklist of endemic speciesfor the flora being assessed that is consistent with current taxon-omy and published in a peer reviewed journal. Unpublished listsand/or inferred endemic concepts can lead to the misapplicationof conservation policy and funds to widespread species with a var-iation in, or invalid, species names. The crucial role a currentchecklist plays in the threatened species assessment is exemplifiedby the 36% reduction in endemic status of Micronesian plant spe-cies from previous estimations made prior to a systematic treat-ment for the region (Costion and Lorence, In press). The qualityof the species assessment utilizing the approach proposed in thisstudy will ultimately rest on the accuracy of applied endemic con-cepts. Often as in this case, these endemic species distributions can

be further defined by known substrate restrictions of particulartaxa (Costion et al., 2009).

3.2. Review of archaeological and botanical literature on savannaorigins

Botanists have long proposed that the extent of savanna vegeta-tion on Micronesia’s islands was restricted before colonization andexpanded due to human activities (Merrill, 1912; Fosberg, 1960,1962; Barrau, 1961). Under this scenario, pockets of grassland wereconfined to the fresh volcanic slopes in the Northern Marianas andpotentially poorly drained soils in other localities. Settlement ofthese largely forested islands was accompanied by burning of thenative vegetation to clear land for agriculture (Barrau, 1961),habitation, and line of sight, and to encourage grass growth forthatching material (Kirch, 1982), among other activities. Ananthropogenic cause for Micronesia’s savannas has substantial sci-entific support largely due to the palaeoenvironmental coring workof Athens and colleagues (Athens and Ward, 2004, 2005; Athenset al., 2004; Athens and Stevenson, 2010).

Other researchers disagree and point out the substantial roomfor misinterpretation of the palaeoenvironmental data (Spriggs

(1) Establish endemic status of species and total area of occupancy. Priority levels(1) < 10 km2, (2) < 500 km2 and (3) < 2,000 km2

(2) Review archaeological and botanical literature for:

(a) Archaeological evidence for anthropogenic disturbance

(b) Theories and evidence for origins of extant vegetation

(a) Locality specific evidence of historical deforestation

(3) Formulate hypothesis for anthropogenic origins of current vegetation type

(4) Confirm origin of vegetation type with:

(b) Taxonomic evidence (e.g. presence of endemic species)

(5) Quantify total extent of historical habitat decline from:

(a) Surveyed archaeological sites (total extent proven)

(b) Existing vegetation maps (total extent inferred)

(6) % inferred decline in original habitat qualifies endemic species as At Risk (! 30%), Threatened (! 50%), or Extremely Threatened (!80%)

(7) Threatened status is globally recognized

Fig. 3. Outline of proposed interdisciplinary species assessment methodology.



and Anderson, 1993; Anderson, 1994) or provide natural causalmechanisms for savanna expansion (Zan and Hunter-Anderson,1987; Hunter-Anderson, 2009). The latter authors present an alter-native geoclimatic model, arguing that Pacific island savannasdeveloped by natural means during the drier climate of the Pleisto-cene and fluctuated in size throughout the Holocene due to climatechanges, edaphic conditions and natural fires. We summarize ourcomplete review (Supplementary Data 3) below:

3.2.1. The anthropogenic model

" There is substantial palynological evidence for a correlationbetween an increase in savanna indicator species and humancolonization of the islands. Prior to human colonization theseindicators are minimal or entirely absent from pollen profiles." Several unknown pollen types recorded in Palau’s sediment

cores from pre-human times rapidly decrease, some disappearentirely, after human colonization. Similar findings are recordedon other Pacific islands." As the savanna indicators, replicated in multiple cores from

widely scattered locations throughout Palau and the Pacific,do not occur simultaneously (Table S1), decline in native forestis interpreted to result from an intense level of human activityrather than being due to a natural phenomenon such as climatechange." Archaeological evidence found in savanna-covered monumen-

tal earthwork complexes (Figs. 4) suggests an anthropogenicorigin of the vegetation type (See Sections 3.3.1 and 3.3.3).

3.2.2. The geoclimatic model

" The presence of charcoal in the fossil record long before possiblehuman colonization suggests natural occurrence of fires. How-ever this argument is taken out of context from the relativeamount and size of the particles occurring prior to colonization.Microscopic charcoal can be in record before people arrived dueto atmospheric transport but is much smaller in size than thecharcoal particles associated with local human activity. Furthermore Athens and Ward (2005) demonstrated that charcoal andnaturally occurring lignite cannot be distinguished in palyno-logical sequences." Soil conditions are suggested to have had a contribution to the

natural occurrence of savannas however there is no geologicalor edaphic evidence to support this claim in Palau." Climate change particularly the El Niño/Southern Oscillation

(ENSO) weather system is invoked to account for a natural ori-gin of savannas. This claim is drawn from trends observed inother distant bioregions and lacks any evidence for a correlation

in timing of the onset of change between the two bioregions nora correlation between recorded changes in climate and increaseof savanna indicators in pollen profiles." The existence of plants that are allegedly endemic to the

savanna vegetation are considered proof for the natural originof savannas. However taxonomic review of these species indi-cates that these species naturally occur in other native habitatsand are not restricted to savannas in Micronesia (See Section3.3.2).

3.3. Hypothesis – Palau’s savannas are anthropogenic

Concurring with previous authorities (Mueller-Dombois andFosberg, 1998; Manner et al., 1999; Athens and Ward, 2004), wehypothesize that Palau’s savanna vegetation is all or nearly allanthropogenic, with pre-human occupation extent being minimal.The immediate implication of this hypothesis is that quantificationof the savanna vegetation will provide a minimum total extent ofhabitat decline since human colonization. If our hypothesis is cor-rect then there should be substantial evidence for ancient and his-toric large-scale deforestation. We should also expect to find few ifany endemic plants restricted to savanna habitat due to the recentorigin of savannas in Micronesia on the evolutionary timescale.Although the alleged existence of savanna endemics has been dis-cussed previously, this is the first time the issue is addressed with asystematic review of the taxonomic evidence.

3.3.1. Locality specific archaeological and historical evidence ofdisturbance

Clearing of Babeldaob’s forest likely began at colonization (be-tween 4300 and 3200 cal yr BP). Most of the very early impactsto land-cover are not detectable in the archaeological record orquantifiable in the palaeorecord. However, construction of oftenmassive earthwork complexes from c. 2400 to 1200 cal yr BP fol-lowed by the formation of more modest step-terraces that contin-ued until several hundred years ago left an indelible mark on thelandscape (Liston, 2007, 2009). A significant amount of ancient ter-racing has been identified and mapped by a combination of pedes-trian field survey and remote sensing methods (Liston, 2012)providing spatial proof for areas that have undergone habitat de-cline in the past.

Babeldaob’s transformation into an agricultural and industriallandscape during the Japanese era (1914–1944) also resulted inwide-spread deforestation (Iida, 2011). The Nanyo-Cho (South SeasGovernment) located its capital on Koror in 1922 and Palau be-came the primary agricultural production center for Japan’s Micro-nesian colonies with plantations producing enough cash crops forexport to Japan and forestry operations exploiting local timberfrom forests for firewood and charcoal. Bauxite strip-mining oper-ations from 1940 to 1944 yielded 369,227 tons of bauxite ore,meeting nearly 10% of Japan’s aluminum consumption (Miyaoka,2010). With Japanese vegetation survey maps and World War IIera aerial photographs, a percentage of these activities were iden-tified and quantified from plantations, settlements, military train-ing grounds and other sizeable enterprises (Liston et al., in press).

3.3.2. Taxonomic evidence – Micronesia’s savanna endemicsPrevious inferences that the occurrence of endemic plant spe-

cies in Micronesia’s savannas is evidence of naturally occurring an-cient savannas were based on poorly understood and confusedtaxonomic regional boundaries and also lacked evidence from fieldinventory of the candidate species. The vast majority of Microne-sia’s endemic plant species, 364 (Costion and Lorence, In press),are restricted to forest habitat with comparatively few endemics(Table 2) potentially constrained to the savannas (see Supplemen-tary Data 4 for detailed review).

Fig. 4. Step-terrace complex in the Ngaremlengui Earthwork District (Photo by J.Liston).



Our review of Palau’s potential savanna endemics suggests thatthey all can be placed in at least one of two categories: (1) speciesoccurring both in savanna and other sun-exposed native habitats,and (2) species with insufficient taxonomic support for use in sci-entific hypotheses. Extensive vegetation surveys conducted onBabeldaob identified a high correlation in species composition be-tween savanna-forest edges and upland ridgeline forests (Costionand Kitalong, 2006). This association, confirmed by a vegetationcluster analysis performed using Two-Way Indicator Species Anal-ysis (Costion, 2007), was attributed to the comparable levels of sunexposure or canopy closure between the two habitats. Other sa-vanna associates not found along upland ridgeline forests areknown to occur on exposed basalt outcrops receiving full sunlight.

Existing data on the taxonomy and distribution of savanna taxasuggest that prior to human settlement, savanna vegetation waspossibly associated with outcrops or the volcanic slopes of theMariana Islands from which they later dispersed to areas clearedof forest by humans. Ample records of Micronesian ‘‘savanna ende-mic’’ plants occurring in sun-exposed habitats other than savannadirectly refute the argument that shade intolerant plants are evi-dence of the antiquity of savannas (Hunter-Anderson, 2009).

Advocates of the conflicting hypotheses agree that savannalands originated during the Pleistocene, very recently from an evo-lutionary perspective. Although Palau is geologically dated toapproximately 30–37 million years old, the onset of climate changein the Pleistocene would not have allowed sufficient time for a di-verse radiation of species to take place. Ultimately, if Palau’s savan-na vegetation had any relictual character then it should persistwithout human intervention. Studies have shown the contrary,that when savanna areas are left alone and not subjected to inten-tional burning, the adjacent forest expands to reclaim the savanna(Supplementary Data 5).

3.3.3. Total extent of habitat decline proven – historical evidenceEvidence for historic terrace complexes occurs on 45.5 km2

(14.2%) of Babeldaob’s volcanic landscape with an estimated min-imum of 18 km2 additional forested sites yet to be mapped (Liston,2012). Therefore at least 63 km2 (20.1%) of Babeldaob was shapedinto earth structures in the past. Of this estimated total, approxi-mately 14 km2 (22%) of terraced land is currently in savanna veg-etation and the remaining 49 km2 (78%) is forested. Theseforested earthwork sites are direct evidence of forest regenerationwhile the savanna sites are evidence of habitat decline withoutrecovery.

Analysis of cartographic data indicates that during the Japaneseera, Babeldaob’s forest declined by 52 km2 largely due to agricul-tural development, strip-mining operations, and military build-up (Iida, 2011). Only about 8 km2 of this historic deforestation oc-

curred on land cleared during ancient earthwork construction (Lis-ton et al., in press).

Babeldaob’s quantifiable change in land-cover from ancienttimes until 2006 is identified to be about 31% (97.6 km2) of the is-land’s volcanic terrain (Liston et al., in press). This calculation in-creases to at least 37% (115.6 km2) when the estimatedundocumented earthwork sites are considered. At least 55% of Bab-eldaob’s current savanna vegetation coincides with the archaeolog-ical and cartographic evidence for human forest clearing directlysupporting an anthropogenic origin. When these quantificationsare recalculated to exclude Babeldaob’s limestone forest and in-clude Palau’s other volcanic islands the total of proven habitat de-cline or fluctuation amounts to 31% of Palau’s volcanic island totalland area. Many other ancient activities (e.g., swidden cultivation)that required extensive land clearance left behind little archaeolog-ical surface evidence in Babeldaob’s transforming landscape.Hence, the extent of original forest habitat decline should be sub-stantially more than the archaeologically quantifiable (Liston et al.,in press).

3.3.4. Total extent of habitat decline inferred – the current vegetationIntact habitat on Palau’s volcanic islands totals to c. 278 km2

(74%) and disturbed habitat equals c. 98 km2 (26%) (SupportingInformation 6). Adding proven land-use history data (3.5.1), thearea of disturbed habitat increases to 147 km2 (39%), when re-for-ested earthworks (49 km2) are subtracted from the ‘intact habitats’category. This provides a more accurate quantification of currentvegetation impacted by human activities in the past, accountingfor possible biodiversity losses that may have occurred when theforest was originally cleared. The inferred total allows for an in-formed application of the precautionary principle consideringextinction trends recorded for other islands throughout the Pacific.A minimum of 13% of forest habitat fluctuation, where original for-est was cleared then subsequently reforested, is directly quantifi-able. As Palau’s savanna-forest edges are known to responddynamically to the fires routinely ignited by humans the actualfluctuation is likely much higher. Although providing the best esti-mate to date of disturbed habitat on Palau’s volcanic islands (39%),these figures are considered a low estimate of vegetation impactedby human activities.

3.4. Assigning threatened status from inferred habitat decline

The data and studies reviewed have enabled calculation of theminimum extent of human altered habitat, where at least 55% ofPalau’s extant savanna is known to be of anthropogenic origin. Evi-dence from the fossil pollen record suggests that it is highly unli-kely that the remaining savanna—that lacking archaeological

Table 2Candidate ‘‘savanna endemics’’ of Micronesia.

Family Species Endemic to Notes

Malvaceae Trichospermum ledermannii Palau Common in disturbed areas and savannas, likely also occurs on exposed rock outcropsMyrtaceae Myrtella bennigseniana Micronesia Shrub occurring in savanna and disturbed areasPhyllanthaceae Phyllanthus palauensis Palau Shrub, not confined to savanna, also occurs on exposed volcanic rock outcrops

Phyllanthus saffordii Marianas Shrub occurring in savannaGlochidion marinanum Micronesia Shrub occurring in savanna and on limestone

Poaceae Ischaemum longisetum Marianas Herb occurring in limestone coastal areas and savannaPanicum palauense Palau Grass occurring in savanna, very poorly known and studied

Rubiaceae Hedyotis aimiriikensis Palau Likely a synonym with H. korrorensis, only known from type collectionHedyotis korrorensis Palau Shrub, not confined to savannas, also occurring on exposed volcanic rock outcropsHedyotis laciniata N. Marianas Shrub confined to Northern Marianas volcanoes in open savannaHedyotis sachetiana Palau Doubtful species, known only from type which lacks flowers. Likely synonym of H. tomentosaHedyotis suborthogona Palau Potentially synonym with H. fruticulosa. Only known from two collections.Hedyotis tomentosa Palau Small shrub, common in savanna but also known from other exposed, and eroded habitats.Morinda pedunculata Palau Shrub common in open savanna, also occurring along savanna edge and along forested ridges.Timonius albus Yap Shrub or small tree; extent of occurrence unknownTimonius subauritus Palau Shrub common in open savanna, also occurring along savanna edge and along forested ridges.



evidence of human disturbance—is naturally occurring. If all savan-na vegetation is considered anthropogenic, then 26% of Palau’s vol-canic terrain is inferred to have undergone substantial habitatdecline without recovery (Supplementary Data 6). When the de-cline in habitat includes areas known to have been re-forested afterhuman disturbance, an additional 13% of Babeldaob’s habitat hasundergone a directly quantifiable historic fluctuation, making theoverall total of inferred native habitat decline in Palau 39%. Thisconservative inference (5b) is strongly supported since the totalextent of proven habitat decline is 31% (5a).

The evidence presented here would qualify all endemic plantsrestricted to forest habitat on Palau’s volcanic terrain for IUCN sta-tus of Vulnerable; a total of 31–39% in historic habitat decline orfluctuation meets the requirements of Criterion A. However,extending the time considered for these changes beyond three gen-erations does not enable the assessment of these species with IUCNCriteria. Our proposed alternative methodology however, removesthis time constraint enabling Palau’s volcanic island restricted en-demic plants to be listed as ‘‘At Risk (Priority Level 2)’’.

3.5. Globally recognizing threatened status of species

We suggest that the neglect of historical evidence beyond threegenerations is a fundamental weak point in the use of IUCN criteriafor assessing plant species in many tropical regions, and concurwith Régnier et al. (2009) and Cardoso et al. (2011) that the IUCNCriteria are more suitable for vertebrates. Most vertebrate groupsare more completely known than plants, benefiting from a largerecologically based literature. The threatened status of both plantand invertebrate species relies primarily on the taxonomic litera-ture, which is sparse in most tropical regions. Cardoso et al.(2011) provide a complete assessment of the limitations of theIUCN system for invertebrates of which many are shared withplants. IUCN criteria are easier to apply for large sized speciesand species with widespread distributions and biases against localendemics, which compromise the vast majority of insect and plantspecies. In the case of Palau, there are known endemics with rangessubstantially smaller then the total documented deforested landarea (Costion et al., 2009). The 115.6 km2 of area proven disturbedby Liston et al. (in press) could have maintained habitat for overhalf the population of some of these endemic species. Even thougha small proportion of this area has subsequently re-forested, loss ofhabitat in the tropics or biodiversity ‘‘hot spots’’ is a more seriousconcern than in temperate areas or ‘‘cold spot’’ regions, which areknown to be more resilient and able to re-forest with minimal orno species loss (Pimm et al., 1995). Palaeocores document uniden-tified fossil pollen types present before Palau was colonized andabsent after the advent of humans (Athens and Ward, 2002,2005). Endemic plant species may have gone extinct and the extentof occurrence of others may have declined substantially in re-sponse to human activities. This could potentially explain someof the disjunct and scattered distributions of some of Palau’s ende-mic plants.

In the Pacific few quantitative studies have yet to be initiatedfor plants and conducting them comprehensively by 2020 is a farfrom realistic target. Data on changes in species abundances inspace and time, considered one of the main shortfalls of applyingIUCN Criteria for insects (Cardoso et al., 2011), is also severely lack-ing for plants. An approach, such as the one suggested in this paper,that considers evidence of population declines over much longerperiods, however, is informative about the status of species underconsideration and can be done utilizing existing data. This infor-mation can then be weighted against the likelihood of extinctionfrom current threats. This method may prove to be useful on otherislands in the Pacific and beyond where researchers and policymakers struggle with the same data limitations. Whilst the method

developed here can be applied to a range of bioregions, it may re-quire modification and adaptation to specific biomes with differ-ences in the availability of data. In many bioregions, however,the data required for the application of this methodology alreadyexists and will only require the necessary collaboration of qualifiedresearchers from the respective disciplines.

Local governments can devise new criteria for assessing threa-tened status of their own species that are more flexible than theIUCN ‘‘one size fits all’’ model. The United States Endangered Spe-cies Act of 1973 is an example of a dynamic system that allows forpetitioning of species for consideration to the list using any avail-able evidence that suggests the species of concern is threatenedwith extinction or is considered ‘‘at risk.’’ Recently, 48 species ofplants endemic to the Hawaiian island of Kaua’i were added tothe US Endangered Species List (US Fish and Wildlife Service,2010). The assessment process for these species utilized a similarecosystem level approach, as in the present study, where an entireecosystem or vegetation type was quantified and classified asthreatened. Such an approach is likely to prove to be more practicalin areas where little quantitative data is available for populationsof species that are clearly vulnerable to extinction and could easilybe adapted to local legislations or tribal laws where appropriate.

4. General discussion

The findings presented in this regional study should not be con-fused with other more global discussions on anthropogenic versusclimate derived savannas or dry forests (Keeley and Rundel, 2005).In other regions, it is established that savannas and dry forests pre-date the occurrence of humans. Australia is a well supported exam-ple where continental-scale aridification followed its separationfrom Antarctica and lead to the evolution of Australia’s iconic scle-rophyll vegetation. Australia’s sclerophyll vegetation has evolvedinto a great diversity of ecotypes ranging from open savannas tomonsoon forests and undoubtedly predates the arrival of humans.Rainforest boundaries in Australia are also known to have ex-panded during warm, wet phases and contracted during glacialmaxima (Hilbert et al., 2007).

Similar climate driven vegetation changes have been reportedin other regions. The Brazilian cerrado has been shown to haveoriginated in a complex interaction of changing climate, naturalfires, and edaphic conditions (Pinheiro and Monteiro, 2011) andits current species assemblage is attributed to the rise and domi-nance of flammable C4 grasses within the past ten million years(Simon et al., 2009). Both the cerrado and Australian sclerophyllplants can been distinguished from the plants in Micronesia’s sa-vanna by their more obvious morphological adaptations to droughttolerance. The savanna vegetation of Southeast Asia has also beendiscussed by many authors and is understood to have been drivenby climate change (Bird et al., 2005) forming a savanna corridor atparticular times. A recent study (Cannon et al., 2009) inferred thatthe current rainforest from the Sunda plate is actually in a refugialstate having been reduced from a more widespread extent at thelast glacial maximum. This opposite trend to that observed in Aus-tralia and the northern hemisphere sheds further doubt and cau-tion towards assuming direct correlations between climatedriven vegetation patterns in Southeast Asia to the west in Micro-nesia without the aid of rigorous climate modeling approachesbased on data derived from there.

The occurrence of lightning during dry periods is known tohave been a significant ecological factor in the formation of sav-annas in Africa (Keeley and Rundel, 2005). Although this causalmechanism has been invoked as a possible explanatory factor inMicronesia (Hunter-Anderson, 2009), no lightning caused fireshave been recorded to date in Palau. The lack of a pronounced



dry season in Palau compared to continental regions likely pre-vents such occurrences.

The available evidence in Palau supports an anthropogeniccause of savanna formation and that this vegetation type mayhave no past analog. Plant species occurring there originate froma variety of other sun exposed habitats together forming a novelspecies assemblage. The emergence and existence of novel spe-cies assemblages is not a new phenomenon but is now a rapidlygrowing global pattern and in many cases poses a serious threatto biodiversity. Methods that can distinguish between novel andindigenous species assemblages as in the present study shouldbe considered a critical component of biodiversity conservation.

A recent study (Rivers et al., 2011) demonstrated how threa-tened status of plant species can be inferred from existing herbar-ium records. Although this method may be useful for manybioregions, it was determined that an average of 15 collection re-cords per species are required to apply the method accurately. Inthe case of Palau, up to 70% of the native flora is known from onlyfive or fewer collections (Costion, 2007). Our method of establish-ing threat may be considered a first step at establishing baselinethreat, specifically in poorly sampled localities, with existing datafor identifying global priorities and progressing towards the GSPC2020 target. It should not however, be considered a replacementfor determining local abundance and/or rarity of the respectivetaxa within their extent of occurrence, which is clearly the secondstep. In the present case, knowledge of all the endemic taxa to Pa-lau’s volcanic islands is uneven. It is known that species previouslythought to be rare or uncommon turned out to be locally wide-spread once their identity was clarified. In addition, several narrowrange endemic species have turned out to be synonyms or invalidspecies. Floristic inventories conducted in remote tropical regionsare also often designed to achieve different outcomes and/or aresubject to uneven taxonomic expertise or interest. Our method en-ables global priorities to be made which should then be followedup with systematic local inventories to assess abundance and rar-ity in addition to taxonomic research to update and/or verify spe-cies concepts. These inventories must be systematic with rigorousquantitative methods and voucher material to account for allunidentified records. This data can then be used to establish aranking system within our criteria, e.g. Rare, Uncommon, Common,Abundant. The full listing would then read: Genus species At Risk,Priority Level 2, Abundant locally.

This three tiered system captures the abundance of species fromboth a global and local perspective and qualifies its threat based onthe percent of its known habitat decline. The first two levels of datashould be considered sufficient for establishing global threat andmeet the requirements for the GSPC target whilst obtaining thethird level of local data can then be prioritized by region basedon threat and priority levels.

We concur with Cardoso et al. (2011). That modifications tothe current IUCN Criteria are required to enable more poorlyknown species to be assessed but propose this to be done withinthe existing IUCN framework. In our case this is simply done bymodifying one of the existing criteria. A clause could be madewhere exceptions to the existing 10 year/3 generation time periodare permitted for certain taxonomic groups or bioregions wheredata is lacking. We do not propose the creation of an entirelynew assessment framework. Gauthier et al. (2010) argue that‘‘priority setting’’ for conservation actions on rare species doesnot equate with ‘‘extinction risk’’ for which the IUCN red list isdesigned to identify however, the IUCN red list is the primarydata source in many countries for establishing such priorities.For IUCN to remain effective to this end it must heed the manyvoices of the scientific community that urge for revisions to ac-count for the severe data gaps and biases within the currentsystem.

The importance of accounting for uncertainty in conservationplanning for species has been recently highlighted (Butchart andBird, 2010; Gillespie et al., 2011) to avoid the risk of ignoringpotentially threatened species. Butchart and Bird (2010) found thatthe highest percentage of data deficient (DD) bird species on theIUCN red list occur in subtropical or tropical lowland moist forests.Our case study in the tropical moist lowland forests of Palau, viv-idly verifies this statistic for plants, with over 70 data deficient spe-cies which qualified as threatened after permitting a minormodification to the IUCN assessment process. If IUCN will not per-mit necessary modifications to the assessment process then thescientific and conservation community at large should seriouslyconsider applying the ‘‘pre-cautionary principle’’ to species thatare listed as data deficient (DD) as suggested by Butchart and Bird(2010) and substantiated here for plants.

5. Conclusions

Despite widespread acknowledgment of the need to accelerateassessments of threatened plants, the process is largely impededby both the IUCN assessment criteria’s high standards and the pau-city of high resolution distribution data for plant species in tropicalregions where the majority of plant diversity occurs. To achieve theGSPC 2020 targets it is essential that either revisions be made tothe existing IUCN criteria that would facilitate a more rapid assess-ment of plant species in these regions, or that the adoption of alter-native plant assessment methodologies are widely recognized. Wepresent one possible solution for the latter option by enabling theuse of historic and archaeological evidence of landscape changeand disturbance. On Palau’s volcanic islands, evidence of ancientand historic land use identifies human activity deforesting a bareminimum of 31% of Palau’s volcanic terrain. Although the direct ef-fects of this deforestation on the endemic plant species are un-known it is certain to have caused reductions in their originalpopulation sizes.

In this study we also addressed a long debated hypothesis onthe anthropogenic origins of Palau’s savanna vegetation using aninterdisciplinary approach. While the palaeoenvironmental recordlends robust support, both substantial quantifiable archaeologicaland cartographic data and the lack of taxonomic evidence for aclear endemic savanna component strongly substantiates the sa-vanna vegetation as primarily anthropogenic. Open savanna typevegetation may have existed in the Northern Marianas, which havea recent volcanic history, prior to human settlement. However, thedata indicates that before colonization, Palau was an essentiallyforested landscape with probably very few pockets of open terrain.At present, there is little support from existing species level datafor the climate driven savanna origin theory in Micronesia.

This study raises questions with major implications for not onlythe Pacific but the field of biodiversity conservation at large. Canthe threatened status of extant populations of plant species be as-sessed accurately when only three generations of their evolution-ary history are considered? What are the effects of ancient andhistoric forest clearance and subsequent revegetation on islandbiodiversity? Should re-vegetated ancient sites whose current hab-itat appears natural be given equal biodiversity value as potentiallyundisturbed habitats? Beyond traditional archaeological field sur-vey, what practical methods are available to identify these anthro-pogenic habitats?

Perhaps most importantly, since palaeoenvironmental andarchaeological investigations capture key information and lessonsabout past land-use practices and human adaptations to changingenvironmental conditions should environmentalists and ecologistsnot be utilizing data from these fields of research to inform currentconservation policies and decisions?



Acknowledgements

We thank the Critical Ecosystem Partnership Fund (CEPF) forfunding this project and all those who contributed ideas and com-ments to the development of the manuscript including James Mill-er, Mike Balick, Art Whistler, Stephen Athens, Yoji Kurata, SatoshiOsawa, Mikiko Ishikawa, Susan Laurance, and Kirk Costion.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.biocon.2011.12.026.

References

Anderson, A., 1994. Paleoenvironmental evidence of island colonisation: a response.Antiquity 68, 845–847.

Athens, J.S., Stevenson, J., 2010. Holocene paleoenvironmental investigations. In:Liston, J. (Ed.), Archaeological Monitoring and Emergency Data Recovery for theCompact Road, Babeldaob Island, Republic of Palau. Historic PreservationInvestigations, Phase III. Volume XII. Draft report prepared for the U.S. ArmyCorps of Engineers, Pacific Ocean Division, Hawai‘i. International ArchaeologicalResearch Institute, Inc., Honolulu.

Athens, J.S., Ward, J.V., 2002. Holocene Paleoenvironmental Investigations onNgerekebesang, Koror, South Babeldaob, and Peleliu Islands, Palau. Prepared forPalau National Communications Corporation, Republic of Palau. InternationalArchaeological Research Institute, Inc., Honolulu.

Athens, J.S., Ward, J.V., 2004. Holocene vegetation, savanna origins and humansettlement of Guam. In: Attenbrow, V., Fullagar, R. (Eds.), A Pacific Odyssey:Archaeology and Anthropology in the Western Pacific. Papers in Honour of JimSpecht. Records of the Australian Museum, Supplement 29, Sydney, pp. 15–30.

Athens, J.S., Ward, J.V., 2005. Palau Compact Road Archaeological Investigations,Babeldaob Island, Republic of Palau; Phase I: Intensive Archaeological Survey;Volume IV: Holocene Paleoenvironment and Landscape Change. Prepared forthe US Army Corps of Engineers, Pacific Ocean Division, Hawai‘i. InternationalArchaeological Research Institute, Inc., Honolulu.

Athens, J.S., Tuggle, H.D., Ward, J.V., Welch, D.J., 2002. Avifaunal extinctions,vegetation change, and polynesian impacts in prehistoric Hawai‘i. Archaeologyin Oceania 37, 57–78.

Athens, J.S., Dega, M.F., Ward, J.V., 2004. Austronesian colonisation of the Marianaislands: the palaeoenvironmental evidence. Indo-Pacific Prehistory AssociationBulletin 24, 21–30.

Baillie, J.E., Bennun, L.A., Brooks, T.M., Butchart, S.H., Chanson, J.S., Cokeliss, Z.,Hilton- Taylor, C., Hoffmann, M., Mace, G.M., Mainka, S.A., Pollock, C.M.,Rodrigues, A.S., Stattersfield, A.J., Stuart, S.N., 2004. 2004 IUCN Red List ofThreatened Species™: A Global Species Assessment. IUCN. The WorldConservation Union. Switzerland and Cambridge, UK.

Barrau, J, 1961. Subsistence Agriculture in Polynesia and Micronesia. B.P. BishopMuseum Bulletin 223.

Bird, M.I., Taylor, D., Hunt, C., 2005. Palaeoenvironments of insular Southeast Asiaduring the Last Glacial Period: A savanna corridor in Sundaland? QuaternaryScience Reviews 24, 2228–2242.

Butchart, S.H.M., Bird, J.P., 2010. Data deficient birds on the IUCN Red List: whatdon’t we know and why does it matter? Biological Conservation 143, 239–247.

Cannon, C.H., Morely, R.J., Bush, A.B.G., 2009. The current refugial rainforests ofSundaland are unrepresentative of their biogeographic past and highlyvulnerable to disturbance. Proceedings of the National Academy of Sciences106, 11188–11193.

Cardoso, P., Borges, P.A.V., Triantis, K.A., Ferrández, M.A., Martín, J.L., 2011. Adaptingthe IUCN Red List criteria for invertebrates. Biological Conservation 144, 2432–2440.

Cole, T.G., Falanruw, M.C., MacLean, C.D., Whitesell, C.D., Ambacher, A.M., 1987.Vegetation survey of the Republic of Palau. Resource Bulletin PSW-22. PacificSouthwest Forest and Range Experiment Station, US Forest Service, USDepartment of Agriculture, University of California Press, Berkeley, CA.

Costion, C., 2007. Floristic diversity and protected area prioritization in Palau,Micronesia. MSc Thesis. Royal Botanic Gardens Edinburgh. Edinburgh, Scotland.

Costion, C., Kitalong, A.H., 2006. Babeldaob Forest Survey 2005. The Republic ofPalau. Belau National Museum Occasional Paper 061007, pp. 1–60.

Costion, C., Lorence, D., In press. The endemic plants of Micronesia: a geographicalchecklist and commentary. Micronesica 42.

Costion, C., Kitalong, A.H., Holm, T., 2009. Plant endemism, rarity, and threat inPalau, Micronesia: a geographical checklist and preliminary red list assessment.Micronesica 41, 131–164.

Fosberg, F.R., 1960. The Vegetation of Micronesia. American Museum of NaturalHistory Bulletin 119.

Fosberg, F.R., 1962. Tropical Pacific grasslands and savanna. In: The Secretariat,Ninth Pacific Science Congress (Ed.), Proceedings of the Ninth Pacific ScienceCongress of the Pacific Science Association (1957), Vol. 4, Botany. Dept. ofScience, Bangkok, Thailand, pp. 118–123.

Gauthier, P., Debussche, M., Thompson, J.D., 2010. Regional priority setting for rarespecies based on a method combining three criteria. Biological Conservation143, 1501–1509.

Gillespie, G.R., Scroggie, M.P., Roberts, J.D., Cogger, H.G., Mahony, M.J., McDonald,K.R., 2011. The influence of uncertainty on conservation assessments:Australian frogs as a case study. Biological Conservation 144, 1516–1525.

Hilbert, D.W., Graham, A., Hopkins, M.S., 2007. Glacial and interglacial refugiawithin a long-term rainforest refugium: the Wet Tropics Bioregion of NEQueensland, Australia. Palaeogeography, Palaeoclimatology, Palaeoecology 251,104–118.

Holdaway, R.N., 1989. New Zealand’s pre-human avifauna and its vulnerability.New Zealand Journal of Ecology 12 (Supplement), 11–25.

Hunter-Anderson, R., 2009. Savanna anthropogenesis in the Mariana Islands:reinterpreting the paleoenvironmental data. Archaeology in Oceania 44, 125–141.

Iida, A., 2011. Research on landscape planning based on watershed analysis: A casestudy on Babeldaob, the Republic of Palau. PhD thesis. Department of UrbanEngineering, University of Tokyo, Tokyo, Japan (in Japanese).

IUCN, 2001. IUCN Red List Categories and Criteria: Version 3.1. IUCN SpeciesSurvival Commission. IUN, Gland, Switzerland and Cambridge, UK.

IUCN, 2010. IUCN Red List of Threatened Species. Version 2010.4. <http://www.iucnredlist.org> (Downloaded 27.10.10).

Jennings, D.E., Rohr, J.R., 2011. A review of the conservation threats to carnivorousplants. Biological Conservation 144, 1356–1363.

Keeley, J.E., Rundel, P.W., 2005. Fire and the Miocene expansion of C4 grasslands.Ecology Letters 8, 683–690.

Kirch, P.V., 1982. The impact of the prehistoric Polynesians on the Hawaiianecosystem. Pacific Science 36, 1–14.

Liston, J., 2007. Archaeological Data Recovery for the Compact Road, BabeldaobIsland, Republic of Palau. Historic Preservation Investigations, Phase II. VolumeV: Lab Analyses, Syntheses, Recommendations. Prepared for the U.S. ArmyCorps of Engineers, Pacific Ocean Division, Hawai‘i. International ArchaeologicalResearch Institute, Inc., Honolulu. [Draft 1999].

Liston, J., 2009. Cultural chronology of earthworks in Palau, Western Micronesia.Archaeology in Oceania 44, 56–73.

Liston, J., 2012. Monumental architecture: Human ecodynamics and socio-politicalevolution of an earthwork landscape on Babeldaob island, Republic of Palau.PhD thesis, Department of Archaeology and Natural History, Research School ofPacific and Asian Studies, Australian National University, Canberra.

Liston, J., Iida, A., Holm, T., in press. Land Use History and Savannah-ForestSuccession on a Micronesian Island: Archaeological and Remote Sensing Data.

Manner, H.I., Mueller-Dombois, D., Rapaport, M., 1999. Terrestrial ecosystems. In:Rapaport, M. (Ed.), The Pacific Islands, Environment and Society. Bess Press,Honolulu, HI.

Merrill, E.D., 1912. Notes on the flora of Manila with special reference to theintroduced element. Philosophical Journal of Science Series C, Botany 7, 145–208.

Miyaoka, N., 2010. Mitsui no Alumi Seiren to Denryoku Jigyo. Kallos Publishing Co.Ltd., Tokyo, Japan (in Japanese).

Mueller-Dombois, D., Fosberg, F.R., 1998. Vegetation of the Tropical Pacific Islands.Springer-Verlag New York, Inc., New York.

Myers, N., Mittermeier, R.A., Mittermeier, C.G., da Fonesca, G., Kent, J., 2000.Biodiversity hotspots for conservation priorities. Nature 403, 853–858.

Pimm, S.L., Russell, G.J., Gittleman, J.L., Brooks, T.M., 1995. The Future ofBiodiversity. Science 269, 347–350.

Pinheiro, M., Monteiro, R., 2011. Contribution to the discussions on the originof the cerrado biome: Brazilian savanna. Brazilian Journal of Biology 70,95–102.

Pippard, H., 2008. The Pacific islands: An analysis of the status of species as listed onthe 2008 IUCN Red List of Threatened Species.™ IUCN Oceania Regional Office.

Porch, N., 2007. Insect fossils from Pacific island prehuman and archaeologicalcontexts: reconstructing Lost ‘Edens’ and human impacts. QuaternaryInternational 167–168 (Supplement), 328.

Porch, N., Prebble, M., Haberle, S., 2007. Invasion Palaeoecology: using the fossilrecord to reconstruct the history and impact of human-moderated biologicalinvasions. Quaternary International 167–168 (Supplement), 327.

Prebble, M., Dowe, J., 2007. Quaternary plant extinctions: a case study of palm(Arecaceae) lineages from oceanic islands. Quaternary International 167–168(Supplement), 330.

Pregill, G.K., Steadman, D.W., 2004. South Pacific iguanas: human impacts and anew species. Journal of Herpetology 38, 15–21.

Régnier, C., Fontaine, B., Bouchet, P., 2009. Not knowing, not recording, notlisting: numerous unnoticed Mollusk extinctions. Conservation Biology 23,1214–1221.

Rivers, M.C., Taylor, L., Brummitt, N.A., Meagher, T.R., Roberts, D.L., Lughadha, E.N.,2011. How many herbarium specimens are needed to detect threatenedspecies? Biological Conservation 144, 2541–2547.

Secretariat of the Convention on Biological Diversity, 2011. Global Strategy for PlantConservation, Updated Strategy 2011–2020, Targets. <http://www.cbd.int/gspc/targets.shtml> (accessed August 2011).

Simon, M.F., Grether, R., De Queiroz, L.P., Skema, C., Pennington, R.T., Hughes, C.E.,2009. Recent assembly of the Cerrado, a neotropical plant diversity hotspot, byin situ evolution of adaptations to fire. Proceedings of the National Academy ofSciences 106, 20359.

Spriggs, M., Anderson, A., 1993. Late colonization of east Polynesia. Antiquity 67,200–217.



Steadman, D., 2006. Extinction and Biogeography of Tropical Pacific Birds.University of Chicago Press, Chicago.

US Fish and Wildlife Service, 2010. Federal Register (Rules and regulations) 75, pp.1–206.

Zan, Y., Hunter-Anderson, R., 1987. On the Origins of the Micronesian ‘Savannahs’:An Anthropological Perspective. In: Paper presented at the Third InternationalSoil Management Workshop on the Management and Utilization of Acid Soils ofOceania, Palau.


Using the ancient past for establishing current threat in poorly inventoried regions

Documents

Transcript of Using the ancient past for establishing current threat in poorly inventoried regions