Introduction
Bengal Florican Houbaropsis bengalensis is a ‘Critically Endangered’ bustard confined to the terai grasslands of India and Nepal (nominotypical bengalensis) and to the seasonally inundated grasslands of southern Indochina (subspecies blandini) (Collar Reference Collar, del Hoyo, Elliott and Sargatal1996, Collar et al. Reference Collar, Andreev, Chan, Crosby, Subramanya and Tobias2001). H. b. bengalensis is almost entirely confined to protected areas which are both highly disjunct, with possibly little or no genetic interchange between many of them, and inappropriately managed for the species; surveys of these areas in years between 1990 and 2007 yielded a total of 280–401 individuals for this taxon (Donald et al. Reference Donald, Collar, Marsden and Pain2010). The species is judged to be extinct in the Mekong Delta of Vietnam and southern Cambodia (Buckton and Safford Reference Buckton and Safford2004, J. C. Eames and J. D. Pilgrim in litt. 2009), so during the breeding season the population of the subspecies blandini is now confined to the floodplain grasslands in and around the Tonle Sap Lake of Cambodia (Donald et al. Reference Donald, Collar, Marsden and Pain2010). These grasslands have been created and maintained by a combination of scrub removal during agricultural expansion, abandonment or fallowing after episodes of low-intensity rice cultivation, burning and extensive livestock-grazing, which probably mimic or substitute ecosystem functions once provided by now extirpated or scarce wild herbivores (Wright et al. Reference Wright, Lake and Dolman2012, Packman et al. Reference Packman, Gray, Collar, Evans, van Zalinge, Virak, Lovett and Dolman2013), and natural fires. Much of this grassland was surveyed during 2005 to 2007; multiplying observed densities of displaying males by aerial photo-derived extent of suitable habitat in 2005, a population of 376 (293–462) displaying males was estimated at accessible sites (Gray et al. Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a). Assuming a similar density, there could also have been around 30 additional displaying males at inaccessible sites; however, available habitat data were already ten years out of date at that stage, so that fewer birds may have been present by 2005. Assuming an equal sex ratio, Cambodia may have held approximately two-thirds of the global population of Bengal Florican.
Early surveys documented a wide range of threats to the floricans, notably hunting, egg collection and changes in agricultural practices (expansion in some areas and abandonment leading to scrub encroachment in others), and conservation actions began in 2000, initially focusing on surveys and community engagement to reduce hunting (Davidson Reference Davidson2004). However, around the time of the 2005 to 2007 census, a major additional threat to the grassland habitat around the Tonle Sap emerged: industrial-scale dry-season rice production for the benefit of large investors, involving allocation of concessions (public land designated for the use of a private entity, for the duration of the agreement) and the construction of large channels, dams and reservoirs in order to catch receding flood waters at the end of the wet season and provide irrigation throughout the dry season (December–June) (Gray et al. Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a, Packman et al. Reference Packman, Gray, Collar, Evans, van Zalinge, Virak, Lovett and Dolman2013).
Grassland conversion was so rapid that in just 27 months between 2005 and 2007 the species may have lost 28% of its 2005 breeding habitat. In response, in 2007 some 349 km2 of floodplain grassland were set aside for biodiversity and livelihood conservation, in the form of six ‘Integrated Farming and Biodiversity Areas’ (IFBAs) under declarations from the respective governors of the provinces of Siem Reap, Kompong Thom and Kompong Chhnang (Gray et al. Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a). In 2010 these were changed to ‘Bengal Florican Conservation Areas’ (BFCAs) under a declaration from the Ministry of Agriculture, Forestry and Fisheries. This increased the level of legal protection, but the boundaries of the protected areas were redrawn to cover only 173 km2 of three floodplain grassland sites, while extending protection to 138 km2 of open forest non-breeding habitat. Meanwhile, however, conversion of floodplain grasslands accelerated, and the area of dry-season rice in the south-eastern part of the floodplain (2,407 km2), containing all BFCAs and 80% of the florican population (Gray et al. Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a), increased from 50 km2 in 2005 to 383 km2 in 2009 (Packman et al. Reference Packman, Gray, Collar, Evans, van Zalinge, Virak, Lovett and Dolman2013) and continues to expand, both outside and inside the conservation areas. This resulted in a net loss of 19% (from 923 km2 to 751 km2) of the grassland habitat in this area between 2005 and 2009 (Packman et al. Reference Packman, Gray, Collar, Evans, van Zalinge, Virak, Lovett and Dolman2013), with losses continuing since.
Between March and May 2012, a new census was conducted to provide an up-to-date assessment of the status of Bengal Florican in Cambodia. In November 2012, a workshop was held in Phnom Penh, Cambodia, to consider the results of this census and to evaluate the most appropriate conservation responses; the key conclusions of this meeting are outlined in the Discussion.
Methods
Being vulnerable to both aerial and terrestrial predators, bustards are generally cryptic, wary birds, and it is only during the breeding season that males become conspicuous (and therefore countable) through their displays and boldly patterned plumages; at this time, in particular, male birds tend to occupy more open, treeless terrain, presumably in order both to render their displays visible over a larger area and to minimise the risk of ambush by predators (Collar Reference Collar, del Hoyo, Elliott and Sargatal1996). Bengal Floricans in Cambodia show a seasonal pattern of movement, migrating to open deciduous dipterocarp forest beyond the floodplain when the Tonle Sap expands during the monsoonal rains (generally May–November) (Packman Reference Packman2011). They move back to breed in grasslands in and around the floodplain as these dry out, with greatest display intensity of males from mid-March to early May (Gray et al. Reference Gray, Chamnan, Borey, Collar and Dolman2007). Like many bustard species, the Bengal Florican has an ‘exploded lek’ breeding system (Gray et al. Reference Gray, Chamnan, Borey, Collar and Dolman2007).
In 2005 a total of 28 grassland sites larger than 10 km2 (considered sufficient to support an exploded lek of > 4 males) were identified from January 2005 aerial photos (for floodplain sites) and 1996 landcover maps derived from aerial photos and, for sites beyond the floodplain, satellite images (JICA 2000). Nineteen of these were surveyed during 2005 to 2007, while the other nine were inaccessible for security or logistical reasons (Gray et al. Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a). In 2012 we resurveyed these 19 sites (as delineated by the 2005 habitat maps) and added one site (Ang Trapeang Thmor, to the far north of the floodplain) where small numbers of floricans have been recorded since at least 2001 (see Table 1 and Figure 1). The nine blocks unvisited during 2005 to 2007 were not surveyed formally in 2012, but we judge their current florican population to be close to zero; six remained off limits but were considered unlikely (based on satellite imagery) to retain grassland, while three were visited, but contained either no remaining suitable habitat (Prasat Bakong and Kandeak, both in Siem Reap province) or limited areas of heavily disturbed grassland (Krakor, in Pursat province).
Survey sites, their province, area and number of survey squares designated, also showing the area declared as Integrated Farming and Biodiversity Areas (IFBAs) in 2007 and still given protection as Bengal Florican Conservation Areas (BFCAs) by 2012.
1 As determined by Gray et al. (Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a) from 2005 aerial photographs
2 Comprising two adjacent IFBAs: Stoung (Kompong Thom) 28 km2, Chikraeng (Siem Reap) 41 km2
3 Comprising two BFCAs: Stoung 28 km2, Chikraeng 46 km2
4 Part included within contiguous Kouk Preah Beung Trea IFBA
5 Additional site for 2012 survey
Map showing estimates of numbers of displaying males at survey sites in 2012. Sites are coded with a letter (see Table 1).
Populations of males within sites were sampled using replicate 1 km2 squares (reflecting the approximate minimum distance between adjacent territories within an exploded lek: Gray et al. Reference Gray, Chamnan, Borey, Collar and Dolman2007, Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a), without consideration of current habitat distribution, to provide a mean density for that site. At each site a 1-km2 grid was overlaid and a sample of squares proportional to site area was selected, aiming to achieve 20% coverage for sites < 100 km2 (with an enforced minimum of eight squares per site for the smallest sites to ensure sufficient replicates for site-specific estimates) and 15% coverage for sites > 100 km2, with a maximum cap of 24 squares per site (Table 1). Squares were surveyed 7 March–5 May 2012, at times of daily peak male display activity in the early morning and late afternoon (06h00–08h30 and 16h00–18h30) (Gray et al. Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a). An observer would remain in the centre of the square for one hour, recording any displaying males observed. A total of 307 squares were identified for sampling. Of these 17 were inaccessible, due to dense scrub or extensive open water (from rice irrigation), and were therefore assumed to hold no floricans (with a zero recorded for the survey). The remaining 290 squares were each surveyed twice and subsequent analysis of density considers the higher of the two counts. This contrasts with the methodology of Gray et al. (Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a), who visited each of 268 survey squares just once (but with two observers, for one hour).
For each site, the total number of displaying males was estimated by multiplying mean density (maximum count per surveyed 1-km2 square) by the site area. In calculating the overall population estimate, to account for unequal survey effort among sites, density estimates from each site were weighted both by survey effort (i.e. proportion of site surveyed) and by the relative contribution of each site to the total study area, following Greenwood and Robinson (Reference Greenwood, Robinson and Sutherland2006) and Gray et al. (Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a). We estimated detectability in 2012, comparing the numbers of floricans located during just the first or just the second round of visits to the maximum count recorded across both visits. To allow direct comparison with the 2005 to 2007 survey, we adjusted the total estimated population by the detectability achieved in a single visit.
Results
The total estimate for displaying male floricans in Cambodia (and thus for Indochina and the subspecies blandini as a whole) for 2012 was calculated as 216 (156–275). Assuming an equal sex ratio suggests a total population estimate of 432 (312–550) individuals; however, this should be considered with caution, as populations of declining species are often male-biased (Donald Reference Donald2007).
A comparison of the 19 accessible sites surveyed by Gray et al. (Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a), holding the vast majority of the estimated population both then and now, reveals a 44% decline in the estimated number of displaying male floricans, from 376 (293–462) during 2005 to 2007 to 211 (152–279) in 2012. In the 2012 survey, single visits detected 64.4% of the maximum numbers of displaying males detected over two visits. We therefore reduced the 2012 population estimate by 35.6%, to allow direct comparison with numbers estimated for the same sites from single visits during the 2005 to 2007 survey; this suggested that the population may have declined by 64%, assuming that detectability was similar between the two surveys. However, repeat visits to 10 squares containing territorial males during 2002–2003 suggested that detectability was consistently high at that time (Davidson Reference Davidson2004), a finding which formed the basis for the assumption that detectability approached 100% during the 2005 to 2007 survey. It is possible that fewer males displayed consistently by 2012 owing to reduced extent of habitat and greater disturbance from agricultural activities. If detectability was better during 2005 to 2007 than in 2012, then the magnitude of population decline may lie between 44% and 64%.
The numbers of displaying males estimated for each site in 2012 are shown in Figure 1, and numbers by site during 2005 to 2007 (from Gray et al. Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a) and 2012 are compared in Table 2. Sampling intensity was designed to allow an overall population estimate, so caution is required when interpreting trends within individual sites. However, of the 19 sites surveyed during both censuses, seven held no floricans in either survey and three appear to have lost their florican population between the two surveys (Mongkol Borei, Kruos Kraom and Chong Doung). In contrast, some sites appear to have maintained numbers (e.g. Stoung–Chikraeng, Sangkae–Kompong Pring, and Kouk Preah Beung Trea).
Estimated numbers of displaying males by site (with 95% CIs) for 2002–2003 (three sites only, from Davidson Reference Davidson2004), 2005 (sites b, d, e) or 2006 to 2007 (16 sites) (from Gray et al. Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a) and 2012. Sites listed in descending order of population size during 2005 to 2007, % change marked in bold show no overlap in CIs.
Kouk Preah Beung Trea, San Kor and Stoung–Chikraeng (all located predominantly within Kompong Thom province) hold by far the largest numbers of floricans (combined, 69% of the total population), an estimated 54, 53 and 43 displaying males respectively, with the next largest population of only 18 at Baray. However, Sangkae–Kompong Pring has the highest density of displaying males (0.88/km2), followed by Bakan (0.63/km2) (both small sites < 25 km2 in the western floodplain) and Stoung–Chikraeng (0.61/km2). Sangkae–Kompong Pring also supported the highest density in 2005 to 2007.
Discussion
The results of the 2012 census reveal a further serious deterioration in the status of the Bengal Florican in Cambodia, to the point where its long-term survival in the country (and hence the survival of the taxon blandini and the entire Indochinese population) is perilous. If numbers continue to be lost at a similar rate to the estimated decline between 2005 to 2007 and 2012, it is possible that blandini would become extinct within 10 years.
Scrub encroachment due to agricultural abandonment accounted for 65% of the grassland loss that occurred between 1995/6 and 2005, mostly within the inner floodplain (Packman et al. Reference Packman, Gray, Collar, Evans, van Zalinge, Virak, Lovett and Dolman2013).The substantial decline in florican numbers at Veal Srangai after 2002–2003 reflects this successional loss of grassland to scrub. Since it emerged in 2004–2005 (Gray et al. Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a), rapid, large-scale conversion of the Tonle Sap grasslands to intensive dry-season rice cultivation has become the most severe threat to the Bengal Florican remaining in Cambodia (Packman et al. Reference Packman, Gray, Collar, Evans, van Zalinge, Virak, Lovett and Dolman2013). Dry-season rice cultivation and associated irrigation structures were responsible for 95% of grassland losses in the south-eastern region of the Tonle Sap floodplain between 2005 and 2009 (Packman et al. Reference Packman, Gray, Collar, Evans, van Zalinge, Virak, Lovett and Dolman2013).
Conservation measures for Bengal Floricans focused initially (2000 onwards) on anti-hunting campaigns and diagnostic ecological research, leading to the establishment of protected areas to prevent habitat loss and a range of other threats. Unfortunately, these protected areas have proved extremely difficult to retain intact. IFBAs were set up at Stoung–Chikraeng, Veal Srangai, Kouk Preah Beung Trea, Chong Doung and Baray in 2006/7. IFBA protection was withdrawn at Kouk Preah Beung Trea in September 2008 and when BFCAs, recognised as conservation areas of national importance, were introduced in 2010 at Stoung–Chikraeng, Chong Doung and for part (79%) of the original Baray IFBA, 57 km2 of Veal Srangai IFBA and 19 km2 of Baray lost their protected status. Consequently, whilst the BFCAs now receive, in theory at least, a higher level of protection than in their previous form as IFBAs, the area of florican breeding grasslands under protection has decreased by 50% (from 349 km2 in 2006/7 to 173 km2 in 2010). Therefore, in 2012 only 22% of the florican population was estimated to occur within BFCAs (20% at Stoung–Chikraeng, 2% at Baray), compared to an estimated 36% within the former IFBA boundaries in 2007 (Gray et al. Reference Gray, Collar, Davidson, Dolman, Evans, Fox, Chamnan, Borey, Kim Hout and van Zalinge2009a). Furthermore, the remaining BFCAs are suffering substantial encroachment, with almost 20% of their area under intensive cultivation in 2010 (Mahood et al. Reference Mahood, Virak, Chamnan and Evans2012).
Stoung–Chikraeng BFCA has benefited from active Community Management Committees (CMCs), who value the traditional, low-intensity livelihoods they gain from the grasslands. This strong local network helps ensure that illegal habitat conversion and hunting of floricans are reported and rapid enforcement action is taken. CMCs have now also been established at Baray and Chong Doung BFCAs. To reinforce the value of conservation to local communities, wildlife-friendly rice and community-based ecotourism that have proved successful in other parts of Cambodia (Clements et al. Reference Clements, John, Nielsen, An, Tan and Milner-Gulland2010) are also being piloted in BFCAs.
Marked population declines have apparently occurred within both protected (e.g. Chong Doung and Baray) and unprotected (e.g. San Kor) sites, in contrast to the population within the formerly protected Kouk Preah Beung Trea (IFBA status in 2006–2008 only), which has apparently remained stable, despite recent extensive dam construction within all four sites. A number of possibilities may contribute to differing population trends among sites. First, satellite-tracking showed that birds from Chong Doung and Baray use an entirely separate non-breeding area with little suitable habitat remaining, compared to those at more northerly floodplain sites (e.g. Stoung–Chikraeng, San Kor, Kruos Kraom), for which suitable non-breeding habitat persists (Packman Reference Packman2011). Second, dam construction does not always result in total loss of grassland: patches of grass may remain between the boundaries of concessions, or even within concessions if dams do not hold water or cultivation fails for some other reason. Third, at sites where rice crops are harvested early, it is possible that floricans may use the resulting stubbles and fallows, although this is unconfirmed. However, the persistence of adults displaying in fragmented habitat impacted by agriculture, such as at Kouk Preah Beung Trea, may be misleading. Although breeding productivity is notoriously hard to determine, these areas are likely to be suboptimal for floricans, with (a) impeded drainage that often results in unsuitable dense wet grassland vegetation (Gray et al. Reference Gray, Tarrant, Chamnan, Collar and Dolman2009b), (b) reductions in both burning and grazing that favour scrub growth, and (c) greater human disturbance. As floricans are highly site-faithful, with high survival rates once adult (Packman Reference Packman2011), these sites may be experiencing a ‘lag’ before they lose their birds. Without knowledge of breeding productivity, such landscapes should not be accepted as an alternative to extensive intact grassland conservation, although they obviously merit further study.
There are further grounds for concern. Home ranges of floricans are large relative to BFCA extent. For example, the dry season average 95% kernel is 31 km2 for males and 43 km2 for females (Packman Reference Packman2011), which, despite considerable overlap of individuals’ home ranges, suggests that the current BFCAs are too small to support a breeding population that is viable in the longer term (e.g. Stoung–Chikraeng BFCA is 75 km2). Furthermore, dam construction along the Mekong River threatens to modify flooding regimes in the Tonle Sap basin, with reduced extent of floodplain habitats and concomitant changes in land use (Arias et al. Reference Arias, Cochrane, Piman, Kummu, Caruso and Killeen2012, Orr et al. Reference Orr, Pittock, Chapagain and Dumaresq2012).
To date, BFCAs have had only partial success, although they represent an important and practical immediate response in the struggle to save the Bengal Florican in Cambodia. However, the 2012 census shows that in its current form this approach is, on its own or at least without further development, insufficient to ensure the long-term survival and recovery of the population. Lessons learned from Stoung–Chikraeng show that, for BFCAs to fulfil their function, greater support for local communities from provincial and national government is needed. Also, the current level of protection of BFCAs through ministerial decree is unlikely to be sufficient to avert the large-scale degazettements and excisions as well as the piecemeal encroachments suffered by the IFBAs, particularly in the face of continued pressure to expand rice irrigation.
Recommendations for Bengal Florican conservation
BFCAs need clearer demarcation, more regular and thorough patrolling, to faster detect encroachment and illegal activities, and better means to resolve these. A key recommendation is that BFCAs should be given the highest level of protection possible under Cambodian law, in the form of a prime-ministerial sub-decree. Although politically complicated, this is legally and practically achievable, and such protection from the country’s highest authority would give the best possible chance for a future for Bengal Florican in Cambodia.
To increase the amount of grassland habitat available within BFCAs, efforts should focus on reclaiming land currently under illegal intensive rice cultivation, and on the removal and burning of encroaching scrub, accompanied by research to determine breeding success of floricans and optimal habitat management techniques. If opportunities arise, particularly if farming is abandoned on areas of poorer soil, current BFCAs should be expanded and additional BFCAs should be established (e.g. at San Kor, Kouk Preah Beung Trea and Sangkae–Kompong Pring).
Although contentious, land acquisition both inside and outside existing BFCAs may also be an effective immediate conservation measure, and preliminary land price surveys and legal analyses have already been conducted in 2009 (ACCB and WCS 2009). Outright purchases will require substantial funds and are likely to be limited in size and number, with a lek of ten males commanding a land price in the low millions of dollars. However, such costs reflect the current high agricultural conversion value of florican habitats, emphasising the urgency of decisive action. Conservation concessions (i.e. cultivable land leased and managed for grassland conservation) may be cheaper, but would be less secure and less permanent than outright land purchase. Either approach will have to be protected from fresh encroachment, requiring political support, community willingness, monitoring and enforcement as for the BFCAs.
Annual population monitoring has now been expanded to San Kor, Kruos Kraom and Kouk Prea Beung Trea, as well as the BFCAs and Veal Srangai, together encompassing 80% of the national population. This fieldwork allows surveillance of habitat conversion across these sites, but monitoring of habitat encroachment at other sites should also be repeated. Whilst 80% of the florican population is located in the south-east quarter of the Tonle Sap, the remaining small populations to the west may hold important genetic diversity and face different levels of threat, so should not be neglected. Breeding areas cannot be considered in isolation, and efforts are needed to secure breeding habitat at sites for which appropriate non-breeding habitat also persists.
Acknowledgements
This work was supported by a grant from the Critical Ecosystem Partnership Fund (a joint initiative of l’Agence Française de Développement, Conservation International, the Global Environment Facility, the Government of Japan, the John D. and Catherine T. MacArthur Foundation and the World Bank; a fundamental goal is to ensure civil society is engaged in biodiversity conservation) and BirdLife International. We are very grateful to Sin Seanglay, Nuth Menghor, Chab Oddam, Sot Van Doeun, Phok Pannha and Hoeun Seanlay for assistance with field data collection. We thank the following for assistance: the Siem Reap and Kompong Thom Provincial Government, Kompong Thom Provincial Forestry Administration and the Community Management Committees of Stoung, Chikraeng and Baray/Chong-Doung BFCAs. We are very grateful to two anonymous referees for detailed comments that improved this manuscript.
