Experimental Test of a Conservation Intervention for a Highly Threatened Waterbird

Transcription

1 The Journal of Wildlife Management 77(8): ; 2013; DOI: /jwmg.605 Management and Conservation Experimental Test of a Conservation Intervention for a Highly Threatened Waterbird HUGH L. WRIGHT, 1 School of Environmental Sciences, University of East Anglia, Norwich Research Park, Norwich NR4 7TJ, United Kingdom NIGEL J. COLLAR, BirdLife International, Wellbrook Court, Girton Road, Cambridge CB3 0NA, United Kingdom; and School of Biological Sciences, University of East Anglia, Norwich Research Park, Norwich NR4 7TJ, United Kingdom IAIN R. LAKE, School of Environmental Sciences, University of East Anglia, Norwich Research Park, Norwich NR4 7TJ, United Kingdom NET NORIN, BirdLife International Cambodia Programme (BirdLife), #9, Street 29, Sangkat Tonle Basac, Chamkarmorn, Phnom Penh, Cambodia ROURS VANN, Wildlife Conservation Society (WCS) Cambodia Program, #21, Street 21, Sangkat Tonle Basac, Chamkarmorn, Phnom Penh, Cambodia SOK KO, WWF-Cambodia (WWF), #21, Street 322, Sangkat Boeung Keng Kang I, Chamkarmorn, Phnom Penh, Cambodia SUM PHEARUN, BirdLife International Cambodia Programme, #9, Street 29, Sangkat Tonle Basac, Chamkarmorn, Phnom Penh, Cambodia; and People Resources and Conservation Foundation (PRCF), Banlung, Ratanakiri, Cambodia PAUL M. DOLMAN, School of Environmental Sciences, University of East Anglia, Norwich Research Park, Norwich NR4 7TJ, United Kingdom ABSTRACT Human exploitation and disturbance often threaten nesting wildlife. Nest guarding, a technique that employs local people to prevent such interference, is being applied to an increasing number of species and sites, particularly in South-East Asia. Although research has begun to assess the costeffectiveness of nest guarding, case control studies are rare and the circumstances in which the schemes are most useful remain unclear. We experimentally tested the effect of nest guarding for the critically endangered white-shouldered ibis (Pseudibis davisoni), a species exploited opportunistically for food and now largely confined to dry forests in Cambodia. We randomly applied guarded and unguarded (control) treatments to 24 and 25 nests, respectively, at a single site over 2 years. Nest guarding had no detectable effect on nest success, with an overall probability of nest success of at guarded and at unguarded nests. Nest monitoring across 4 study sites over 3 breeding seasons found a combination of natural predation, weather, and anthropogenic activities (robbery and vandalism) responsible for nest failure, although causes of failure remained unknown at 58% of nests. Nest guarding itself increased nest destruction at 1 site, indicating that this intervention needs cautious implementation if only a small proportion of the local community gains benefit. Comparison with other studies suggests that nest guarding effectiveness may be context-specific and differ between species that are exploited opportunistically, such as white-shouldered ibis, and those routinely targeted for trade. Ó 2013 The Wildlife Society. KEY WORDS Cambodia, conservation effectiveness, nest poaching, nest predation, nest protection, nest robbery, Pseudibis davisoni, white-shouldered ibis. Improving nest success is a fundamental conservation measure for many threatened species, including birds (Bell and Merton 2002, Jones 2004) and reptiles, particularly turtles (Spotila 2004). Nests fail for a variety of reasons including natural predation, starvation, adverse weather, human disturbance, and exploitation for food or trade (the effects of which may be substantial; Wright et al. 2001, Tomillo et al. 2008). Nest exploitation and destruction can be mitigated through various interventions, including conservation payment schemes, often for nest guarding (Clements et al. 2010, Niesten and Gjertsen 2010); awareness campaigns (Herrera and Hennessey 2007, Barré et al. 2010); harvesting quotas (Hobbs 2004); law Received: 5 September 2012; Accepted: 10 June 2013 Published: 16 September hughlewiswright@gmail.com enforcement (Cahill et al. 2006); or a combination of these (Boussekey 2000). Nest guarding schemes employ local people to deter human interference at nests. Guard salaries provide an incentive to report a nest site and ensure its success, rather than harvesting its contents. The approach is becoming popular, particularly for the protection of sea turtle colonies (Ferraro and Gjertsen 2009) and, in South-East Asia, for the protection of waterbirds (Clements et al. 2010, Sok et al. 2012), parrots (Widmann and Widmann 2008), raptors (Prawiradilaga 2006, Salvador and Ibanez 2006), and a bustard (Packman 2011). These species are vulnerable because of their conspicuousness, value, or mere proximity to rural communities depending on natural resources. In Cambodia, for example, 12 threatened birds and 1 river turtle species are currently protected by nest guarding (Clements et al. 2010, Packman 2011, Sok et al. 2012) The Journal of Wildlife Management 77(8)

2 Evaluating the effectiveness of management interventions is important to ensure that they achieve desired goals; do not inadvertently increase problems, for example, disturbanceinduced nest failure associated with nest searching and guard presence; and represent efficient use of resources and time (Sutherland et al. 2004, Ferraro and Pattanayak 2006). Although studies have begun to assess the effectiveness of nest guarding (e.g., Ferraro and Gjertsen 2009, Clements et al. 2013) this intervention is very rarely tested experimentally. Unlike other nest protection interventions (e.g., Kragten et al. 2008, Keo et al. 2009), nest guarding schemes are typically implemented across all monitored nests, leaving no unprotected nests as a control treatment; evaluation has therefore frequently depended on population trend data that can be confounded by other factors, such as weather, fluctuation in predator populations, or implementation of other conservation activities. Recent studies of nest guarding effectiveness in Cambodia have shown contrasting results. Sok et al. (2012) found little effect with 3 waterbird species, although no control was included. Clements et al. (2013) used a quasi-experimental method that matched guarded nests in protected areas with unguarded nests in unprotected areas, suggesting that nest guarding increased nest success for 2 waterbird species; however, this was possibly confounded by other conservation interventions and their effect on community behavior. Clearly, the issue requires further study. The critically endangered white-shouldered ibis (Pseudibis davisoni) is one of South-East Asia s most threatened waterbirds (Tordoff et al. 2005). Using this as a model species, we tested the independent effect of nest guarding using a case control design the first randomized experimental test of this intervention. As the number and scale of nest guarding schemes are anticipated to increase, this study provides conservationists with practicable new evidence regarding this intervention s effectiveness. Whether nest guarding is useful depends on the relative importance of anthropogenic (disturbance or exploitation) to other (natural) causes of nest failure. Therefore, we also monitored the causes of nest failure across 4 ibis subpopulations in Cambodia and discuss their implications for conservation. STUDY AREA Western Siem Pang Important Bird Area (Seng et al. 2003) in Stung Treng province, northern Cambodia ( N E) was the principal study site and location of the nest guarding experiment. During fieldwork in , this 138,000-ha lowland forest landscape held the largest known population of white-shouldered ibis, at least 346 birds (Wright et al. 2013b) along with approximately 11,000 people in small settlements concentrated in the center and east of the site (Commune Database Version 6.0, Ministry of Planning, Phnom Penh). Small-scale conservation action had taken place continuously since 2003 with 4 staff dedicated to waterbird interventions and monitoring (BirdLife International 2009) and law enforcement carried out by a member of the local Forestry Administration. Incidence and causes of nest failure were studied at this and 3 additional sites: Kulen Promtep Wildlife Sanctuary, Preah Vihear province ( N E); Mekong Flooded Forest (henceforth Mekong), following the Mekong River between Kratie and Stung Treng towns ( N E); and Lomphat Wildlife Sanctuary, Ratanakiri and Mondulkiri provinces ( N E). These sites held minimum populations of 43, 124, and 278 white-shouldered ibises, respectively (Wright et al. 2013b). Nest guarding of large waterbirds had been implemented at Kulen Promtep (Clements et al. 2013) and the Mekong (Sok et al. 2012), although nests of white-shouldered ibises were only guarded at the latter. All 4 study sites contained deciduous dipterocarp forest with patches of semi-evergreen and mixed deciduous forest, grassland, and rice paddies; extensive braided river channels also occurred at the Mekong. METHODS Field Methods White-shouldered ibis historically occurred across Indochina, but in the twentieth century the population severely contracted so that 89 97% of the world s remaining 927 1,062 birds are now found in Cambodia (Wright et al. 2013b). The species shares its dry forest habitat with human communities dependent on natural resources and, although not valued for trade, is exploited opportunistically for food (Sok et al. 2012; H. L. Wright, University of East Anglia, personal observation). It is a solitary, dry-season breeder (Dec May; Wright et al. 2013a), building nests in tree canopies typically m above the ground. Nesting most frequently occurs in open deciduous dipterocarp forest, forest remnants, isolated trees at rice fallows (Clements et al. 2013; H. L. Wright, unpublished data), or in seasonally flooded forest along large rivers (Sutrisno et al. 2009, Sok et al. 2012). Nests are frequently occupied in successive years (H. L. Wright, unpublished data). We located and monitored nests for 3 breeding seasons ( ) at Western Siem Pang and Kulen Promtep, 2 seasons at the Mekong ( and ), and 1 season at Lomphat ( ). Nest sample sizes were constrained by the scarcity of the study species and few known nests prior to study inception. Nest reward schemes, applied at all sites, overcame this by providing a small cash incentive for local people to report nests. We conducted additional active searching at old nest sites and new localities where ibis pairs were seen regularly. Searches were systematic in Western Siem Pang, where 4 staff worked full time. As nest sites became known to the staff, the contribution of reward-scheme informants fell from 91% of nests in to 40% in each of the subsequent 2 breeding seasons. Staff search effort was less intensive and often opportunistic at the other 3 sites, reflecting lower capacity and/or other management priorities. Local people contributed 67% of nest finds at Kulen Promtep, 89% at Lomphat, and 100% at the Mekong across all years. Differences in white-shouldered ibis density and knowledge of nest locations resulted in contrasting nest sample sizes across study sites, and may Wright et al. Nest Guarding Effectiveness 1611

3 have reduced the accuracy of nest failure estimates at the 3 additional sites. The distances from nests to the nearest settlement did not differ (F 3,96 ¼ 1.79, P ¼ 0.154) among the 4 study sites, indicating similar proximity to people. We trained field staff to monitor nest activity and overall outcome consistently at every site. Monitoring frequency depended on site capacity; visits were typically every 5 7 days at Western Siem Pang, every 7 days at Kulen Promtep, and every 3 14 days at the Mekong and at Lomphat. Nest guards were also trained to monitor nests twice daily and their records were corroborated by field staff observations made during their regular monitoring visits. The position of nests in high or inaccessible parts of the tree canopy precluded direct observation of nest cups, so observers monitored nests from the ground, waiting until movement of adults or chicks was sufficient to determine nest status. Monitoring visits typically lasted minutes and, if no ibis activity was observed after an hour, staff searched under the nest for evidence of anthropogenic or natural causes of nest failure. Failure was confirmed by a lack of nest activity during 2 subsequent visits or, at well-advanced nests, by finding all nestlings dead. For purposes of analysis, we determined causes of failure from tangible evidence only and in circumstances of near or absolute certainty. We considered the cause of failure unknown in cases of scant evidence or in cases of subjective assessment by observers. Anthropogenic failures were indicated by climbing equipment or felling of the nest tree, but other forms of disturbance and use of slingshots were often undetectable. We accepted reports of nest destruction when based on multiple sources or admissions by those responsible. We only recorded natural predation when the event was actually observed. We inferred the impact of high winds with medium-high probability, using knowledge of recent weather and likely susceptibility given the nest s location in the tree canopy. We assumed premature flight, triggered by unknown causes, when near-fledged chicks were found dead beneath the nest with no evidence of predator damage. We recorded partial brood loss opportunistically but may have underestimated its prevalence, as initial clutch size could not be determined by ground-based observations. The degree to which different causes of nest failure were detected or under-recorded may have varied slightly among sites, owing to differences in capacity and frequency of nest visitation. We treat the attributed causes of nest failure as indicative of potential contributory factors rather than as an accurate measure of their relative importance; natural predation and human disturbance were the most likely to be undetected and were potentially responsible for failures where the cause was unknown. We experimentally implemented nest guarding in the and breeding seasons at Western Siem Pang, randomly applying guarded and unguarded (control) treatments to 24 and 25 nests, respectively, and recruiting guards from local communities. Guards discouraged illegal exploitation or disturbance by threatening to report the perpetrators to the local Forestry Administration, but did not intervene in natural events such as predation. Guards and field staff remained concealed and at least 100 m from the nest to avoid becoming a source of disturbance to nesting birds, or a deterrent to potential predators. We protected 14 nests using a single guard at each; guards started work within 1 3 days of nest discovery (depending on availability), were present during daylight hours, and paid US $ 3.75 per day. Site inaccessibility and/or limited transport availability dictated that guards had to camp in the vicinity of the other 10 guarded nests; these had 2 guards to cover the logistics of camping and guarding (each paid US$ 4.25 per day), and may therefore have received greater protection than single-guard nests. Camps were typically >150 m from nests to prevent causing additional disturbance. Distance of nests to nearest settlement was slightly less at guarded (range km from settlement, mean 3.7 km 2.7 SD) than unguarded (range km, mean 5.6 km 3.5 SD) nests (t 41 ¼ 2.06, P ¼ 0.046). Analysis of Nest Survival To determine the effectiveness of guarding and the predictors of nest failure, we modeled Western Siem Pang data from the 2009 to 2010 and 2010 to 2011 breeding seasons. Data quality was sufficient to model nesting stages separately for 1) the combined incubation and chickbrooding stage, when the nest was almost constantly attended by at least 1 adult, and 2) the late-nestling stage, when both adults stopped sitting or crouching over chicks and were often absent together. We chose these stages for 2 reasons. First, hatching date at some nests was not reliably determined from ground-based observations until chicks were large enough to be visible or adult behavior changed, so that nests failing close to this date could not reliably be assigned as an egg- or chick-stage failure. Second, we assume that failure may be affected by chick size, adult ibis presence at nests, and frequency of provisioning. We considered nest outcome in logistic regression models to predict daily failure rate (DFR), including the number of exposure days (that the nest was active and monitored) as the number of binomial trials (Aebischer 1999). Daily failure rate encompassed all forms of nest failure (anthropogenic and natural). In the first round of modeling, we tested the effects of guarding (guarded and unguarded nests) and breeding season ( and ) on DFR in each nesting stage. In the second round, we undertook model selection for guarding, breeding season, and distance to settlement (square-root transformed to reduce leverage) for the incubation and brooding stage. Alternative models were evaluated by Akaike s Information Criterion corrected for small sample size (AIC c ). We did not undertake model selection for the late-nestling stage, as only 1 failure occurred. During preliminary analysis, we examined nesting date measured as the number of days from onset of breeding (the first observation of incubation across all nests in that year) to the completion of brooding (or failure if earlier) at each nest but this was a poor predictor and was not considered further. To indicate the relative importance of variables, we used model-averaged parameter estimates and change in model AIC c when terms were iteratively dropped 1612 The Journal of Wildlife Management 77(8)

4 from the best model (Burnham and Anderson 2002); an increase in AIC of 2 units indicates strong support. We calculated overall probability of nest success using estimated DFR. To determine what magnitude of effect we would be able to detect for nest guarding, given our relatively small sample size and the observed variability in nest success, we undertook a sensitivity analysis. We re-sampled the nest data, randomly allocating nest outcomes (and associated exposure days) to a dummy pseudo-treatment nominally representing groups of 24 guarded and 25 unguarded nests (matching the experiment); we repeated this process 10 times. For each of the 10 iterations, we calculated overall nest success (as above) for each pseudo-treatment group and examined the confidence intervals (CI) around this estimate. We then calculated the mean CI across the 10 iterations. We estimated how large the effect of nest guarding needed to be for us to detect it (at a significance threshold a ¼ 0.05) by adding half the mean CI of 1 pseudo-treatment group to half the mean CI of the other (following Cumming and Fidler 2005). Finally, we re-expressed this minimum detectable effect size as the percentage difference in overall nest success between guarded and unguarded nests, using the estimated nest success of experimentally unguarded nests at the incubation and brooding stage. To compare nest failure prior to and during the nest guarding experiment, we pooled incubation and brooding stage data from the 2009 to 2011 seasons and compared them in logistic regression models with nests from the 2008 to 2009 season (when all nests were unguarded); models included terms for time period, and both time period and guarding (guarded and unguarded nests). RESULTS We monitored 99 white-shouldered ibis nests across the 4 study sites over 3 breeding seasons, with 12, 20, and 29 nests at Western Siem Pang (the principal site) in , , and , respectively. The probability of overall nest success varied between study sites (Table S1 and Fig. S1 available online at and was an estimated 61% at Western Siem Pang (pooling across seasons and guarded and unguarded nests; 95% CI ¼ 49 76%). Thirty-three of the 99 nests failed, but causes of failure remained unknown at 19 (58%) of these. Anthropogenic factors accounted for at least 9 (27%) failures, involving nest robbery (4 nests) and, at the Mekong, destruction by local people (5 nests) envious of the financial benefits received by nest guards. Strong winds were probably responsible for 3 failures (9%). Premature flight by near-fledged chicks caused 1 failure and at least 1 partial brood loss, although what triggered chicks to leave the nest remained unknown. We confirmed natural predation for 1 nest failure, when a southern jungle crow (Corvus macrorhynchos) removed all eggs of a clutch in the absence of adult ibises, and 1 partial loss of a further brood, when this species predated a newly hatched chick. Nocturnal natural predation (e.g., by mammals) could not be detected using our methodology and could have contributed to failures where the cause was unknown. The nest guarding experiment comprised 49 nests over 2 breeding seasons in Western Siem Pang and nest guard salary payments totaled US$ 5,903. We observed only 1 failure in the late-nestling stage (model AIC c ¼ 13.39), resulting in lower DFR (over both breeding seasons) than in the incubation and brooding stage (model AIC c ¼ 76.19; Table 1). Subsequent analysis considers the incubation and brooding stage, during which estimated DFR was similar between nests with and without the guarding treatment (ratio of guarded relative to unguarded nests [b] ¼ 0.25, 95% CI 1.14) and between breeding seasons (b ¼ 1.12, CI 1.52, in relative to ; Table 1, Fig. 1). Overall success during incubation and brooding was only 4.5% greater at guarded than unguarded nests in , and 14.4% in , compared to a minimum detectable effect of 33.5% (for a ¼ 0.05) given the study s sample size. Failure rate also did not differ (in a univariate model) with level of nest protection (b ¼ 0.14, CI 1.70, for nests protected by 2 guardians relative to nests protected by 1). We found no difference in DFR between time periods during and prior to the guarding experiment (b ¼ 0.73, CI 1.04, for relative to ); with guarding also included in this model, we found no effect of time period and no difference between Table 1. Estimates of daily failure rate (DFR) and probability of overall success of white-shouldered ibis nests at Western Siem Pang, northern Cambodia, We modeled incubation and brooding (combined) versus late-nestling stages separately, each containing terms for breeding season and guarding (guarded and unguarded). Neither breeding season nor guarding substantially improved model fit based on changes in the corrected Akaike s Information Criterion (DAIC c ) when the term was removed from the model (negative values indicate an improvement in model fit without the terms). Nest stage Breeding season a Guarding b Nests Exposure days Failures DFR DFR 95% CI Overall nest success Nest success 95% CI Incubation and brooding Guarded Unguarded Guarded Unguarded Late-nestling Guarded Unguarded Guarded Unguarded a DAIC c ¼ during incubation and brooding and during the late-nestling period. b DAIC c ¼ during incubation and brooding and during the late-nestling period. Wright et al. Nest Guarding Effectiveness 1613

5 Figure 1. Daily failure rates of guarded and unguarded white-shouldered ibis nests in northern Cambodia, , during the incubation and brooding stage, by breeding season. We estimated daily failure rates using a binomial logistic regression model of nest outcome data from Western Siem Pang; error bars indicate standard errors. guarded and unguarded treatments (b ¼ 0.64, CI 1.20, at guarded relative to unguarded nests). We identified 3 best-fitting models of nest failure in the incubation and brooding stage, as 2 models fell within 2 AIC c units of the most-supported model (Table 2). However, breeding season and guarding received no support following model averaging, and distance to settlement was only weakly supported; removing distance to settlement and breeding season from the best model increased model AIC c by 1.95 and 0.99, respectively. Model parameters indicated that DFR was greater with increasing distance to settlement (Fig. 2). The best model (model AIC c ¼ 72.12) predicted a reduction in probability of overall nest success from 0.94 at 1 km from settlement to 0.69 at 10 km from settlement in , and from 0.81 to 0.27 over this distance in Again, we found similar DFR among guarded and unguarded nests in models that included the distance to settlement term. DISCUSSION Causes of Nest Failure White-shouldered ibis nest failures were caused by human exploitation, natural predation, and high winds, problems Figure 2. Daily failure rates (DFR) of white-shouldered ibis nests in northern Cambodia, , during the incubation and brooding stage, in relation to distance to settlement and breeding season. Breeding season comprises (solid line) and (dashed line). We predicted DFR using the best-fitting binomial logistic regression model of nest failure at Western Siem Pang. that also affect nesting giant ibis (Thaumatibis gigantea; Keo et al. 2009) and lesser and greater adjutants (Leptoptilos javanicus and L. dubius) in Cambodia (Sok et al. 2012, Clements et al. 2013). Quantifying the relative importance of these causes of failure is not possible, as natural predation and human disturbance may have been disproportionately undetected. Furthermore, reward schemes could have ameliorated human impacts by providing an incentive not to disturb nests; this was most likely at Western Siem Pang and Kulen Promtep where schemes were applied for longest and staff capacity was good, creating relatively high local awareness. Variable implementation of nest reward schemes and other conservation activities may explain differences in nest success among study sites (see available online at www. onlinelibrary.wiley.com). Although failures were anthropogenic in more than a quarter of cases (64% of known-cause failures), more than half of these were provoked by conservation action itself, as nest guarding met with local resentment at the Mekong. Natural predation caused failure of at least 1 nest and brood reduction at another, but is likely to have caused other, undetected failures also. Fewer nest failures occurred in the late-nestling stage, perhaps because the chicks were too large to be predated or too advanced to be abandoned by Table 2. Multi-model inference and model averaging of nest failure models using white-shouldered ibis nest data from Western Siem Pang, northern Cambodia, We provide corrected Akaike s Information Criteria (AIC c ) and Akaike weights (w i ) for each candidate nest failure model. We calculated model-averaged parameter estimates (b) from all candidate models. A black dot indicates inclusion of the variable in the model. a Model number Guarding Distance to settlement Breeding season AIC c ~AIC c b b % CI to to a Difference in AIC c from that of the best model. b Model-averaged parameter estimates (b) and confidence intervals using unconditional standard errors. w i 1614 The Journal of Wildlife Management 77(8)

6 disturbance-wary parents. Given, however, that humans are more likely to exploit nest contents at the late-nestling stage than at any other period (owing to greater conspicuousness in the nest and greater food value of chicks), higher failure during incubation and brooding suggests natural predation may be a more prevalent cause of nest failure than anthropogenic disturbance or exploitation, at least at Western Siem Pang. Further research should assess the sources and levels of natural predation on ibis nests and the impact of human disturbance particularly flushing adult ibis from nests on their susceptibility to predation. Natural predation may also explain the positive relationship between nest failure and distance to settlement. Predators such as civets (Viverridae) and yellow-throated marten (Martes flavigula) are likely to be more abundant in remote parts of forests owing to strong hunting pressure closer to villages, largely for trade (Srikosamatara et al. 1992). The estimated 67% decline in overall nest success from 1 km to 10 km from settlement ( ) may relate to greater mammalian predation at remote nests; such predation also occurs at giant ibis nests in scarcely populated Cambodian dry forests (Keo et al. 2009). By harvesting these predators, humans may have indirectly protected nests close to villages; nevertheless, these conclusions are provisional, as distance to settlement was only weakly supported in models of nest failure. Nest Guarding Effectiveness This study reports the first randomized experimental test of nest guarding, using white-shouldered ibis as a model species. We found little evidence that nest guarding was effective for this species at Western Siem Pang, as daily failure rates did not differ between guarded and unguarded nests. Although we cannot unequivocally conclude a null effect of guarding, the failure to improve nest success by at least a third (the minimum detectable effect for our relatively small sample) calls into question the cost-effectiveness of the intervention for this species, at this site. Guard salaries were equivalent to US$ 246 per nest, indicating the substantial finance required if nest guarding were to be applied to a large proportion of the dispersed breeding population. The null effect is unlikely to be a result of ineffective protection, as guards (present during all daylight hours) were regularly checked on unannounced visits and were seen to intercept passers-by successfully, suggesting that they would have prevented actual cases of human interference. Rather, the result provides another indication that natural predation, not human exploitation, may be the greater threat at Western Siem Pang; attaching plastic baffles to nest trees to deter mammalian predators (Keo et al. 2009) could be a costeffective alternative to guarding here. This study s results differ from those of Clements et al. (2013), who found that guarded lesser adjutant and sarus crane (Grus antigone) nests had substantially higher success rates than unguarded nests. Although adjutant and crane nests are routinely targeted for trade, white-shouldered ibis nests are exploited only opportunistically and for consumption (Sok et al. 2012; H. L. Wright, personal observation); different results may therefore relate to different magnitudes of exploitation threat, with nest guarding effective at nests of traded species but having little impact at nests of lower-value species. However, Clements et al. (2013) contrasted guarded nests in protected areas with unguarded nests in unprotected areas, so that the apparent positive effect of nest guarding may, in part, also reflect changes in local attitude and behaviors brought about by other interventions, such as community-based ecotourism, an agri-environmental scheme, and law enforcement. Empirical tests of conservation interventions face numerous methodological challenges (Ferraro and Pattanayak 2006). This study tested nest guarding at a single site, enabling an assessment of its independent effect, but local awareness of the intervention could have potentially discouraged exploitation at all nests, guarded or unguarded. Although this spillover effect (Pattanayak et al. 2010) cannot be ruled out, DFR did not differ between seasons during and prior to the nest guarding experiment, suggesting that nest survival was not uniformly improved in this way. An alternative method is to apply intervention and control treatments at separate sites, using statistical approaches to control for confounding factors (Ferraro and Pattanayak 2006); however, other conservation measures may conflate with the tested intervention if they occur concurrently and not in controls. In reality, it can be difficult to identify study communities unaffected by some form of conservation action or sites not confounded by other activities. The presence of a research team for several years may itself contribute to local awareness and change local behaviors; analysts must be wary of such constraints when evaluating management actions. The envy-driven destruction of nests at the Mekong highlights the potential for nest guarding to increase nest failure inadvertently. Nest guarding programs rewarding only a small proportion of the local community require careful implementation, as distributive inequality may create local discontent and undermine the success of payment schemes (Sommerville et al. 2010). Conservation activity has only recently begun at the Mekong and low conservation awareness, combined with poor guard diligence, could have contributed to the destructive action taken by some local people (Sok et al. 2012). Improving community engagement measures and guard payment structures (e.g., making payments conditional to nest outcomes) may improve awareness, address perceptions of unfairness, and protect nests from added destruction (Sok et al. 2012). MANAGEMENT IMPLICATIONS With nest guarding found to be ineffective in this study but effective in another (Clements et al. 2013), the value of this intervention may be context-specific. Guarding may be most valuable for routinely targeted (and traded) waterbirds that face a greater threat from nest exploitation than the white-shouldered ibis. Nevertheless, the severity of opportunistic exploitation may vary between sites (plausibly influenced by the history of conservation activity and degree of local conservation awareness) and therefore the effectiveness of guarding species of low trade value deserves further evaluation in a broader range of contexts. Wright et al. Nest Guarding Effectiveness 1615

7 Conservation programs should therefore continue to monitor the effect of nest guarding schemes, paying careful attention to local disquiet over guard payments, and applying a control treatment of unprotected nests wherever possible. ACKNOWLEDGMENTS Special thanks to Lourn Bun Paeng, Mem Mai, and Sok Mab for their dedicated work at Western Siem Pang. We also thank: Bou Vorsak and J. Eames (BirdLife); H. Rainey and T. Clements (Wildlife Conservation Society); M. Grindley (People Resources and Conservation Foundation); G. Congdon and R. Zanre (World Wildlife Fund), N. Butcher and A. Asque (Royal Society for the Protection of Birds), and D. Buckingham (Oriental Bird Club). We are grateful to Men Phymean of Forestry Administration and H. E. Chay Samith of the General Department of Administration for Nature Conservation and Protection, Ministry of Environment of Cambodia, and all the local people and field staff who participated in nest finding, guarding and monitoring. For funding we thank the Angkor Centre for Conservation of Biodiversity, part of the Zoological Society for the Conservation of Species and Populations; Critical Ecosystem Partnership Fund (CEPF); British Ornithologists Union; Natural Environment Research Council; Economic and Social Research Council; Mohammed bin Zayed Species Conservation Fund; Oriental Bird Club; Rufford Small Grants Foundation; and Giant Ibis Transport. CEPF is a joint initiative of l Agence Française de Développement, Conservation International, Global Environment Facility, Government of Japan, MacArthur Foundation, and World Bank. LITERATURE CITED Aebischer, N. J Multi-way comparisons and generalized linear models of nest success: extensions of the Mayfield method. Bird Study 46: Barré, N., J. Theuerkauf, L. Verfaille, P. Primot, and M. Saoumoé Exponential population increase in the endangered Ouvéa parakeet (Eunymphicus uvaeensis) after community-based protection from nest poaching. Journal of Ornithology 151: Bell, B. D., and D. V. Merton Critically endangered bird populations and their management. Pages in K. Norris, editor. Conserving bird biodiversity. Cambridge University Press, Cambridge, United Kingdom. BirdLife International Strengthening local conservation groups in Cambodia to save critically endangered species. Babbler 30:7 9. Boussekey, M An integrated approach to conservation of the Philippine or red-vented cockatoo. International Zoo Yearbook 37: Burnham, K. P., and D. R. Anderson Model selection and multimodel inference: a practical information-theoretic approach. Second edition. Springer-Verlag, London, United Kingdom. Cahill, A. J., J. S. Walker, and S. J. Marsden Recovery within a population of the critically endangered citron-crested cockatoo Cacatua sulphurea citrinocristata in Indonesia after 10 years of international trade control. Oryx 40: Clements, T., A. John, K. Nielsen, D. An, S. Tan, and E. J. Milner-Gulland Payments for biodiversity conservation in the context of weak institutions: comparison of three programs from Cambodia. Ecological Economics 69: Clements, T., H. J. Rainey, D. An, V. Rours, S. Tan, S. Thong, W. J. Sutherland, and E. J. Milner-Gulland An evaluation of the effectiveness of a direct payment for biodiversity conservation: the Bird Nest Protection Program in the Northern Plains of Cambodia. Biological Conservation 157: Cumming, G., and F. Fidler Interval estimates for statistical communication: problems and possible solutions. Session III in Proceedings of the IASE Satellite Conference on Statistics Education and the Communication of Statistics. International Association for Statistical Education. < Accessed 29 May Ferraro, P. J., and H. Gjertsen A global review of incentive payments for sea turtle conservation. Chelonian Conservation and Biology 8: Ferraro, P. J., and S. K. Pattanayak Money for nothing? A call for empirical evaluation of biodiversity conservation investments. PLoS Biology 4:e105. Herrera, M., and B. Hennessey Quantifying the illegal parrot trade in Santa Cruz de la Sierra, Bolivia, with emphasis on threatened species. Bird Conservation International 17: Hobbs, J. J Problems in the harvest of edible birds nests in Sarawak and Sabah, Malaysian Borneo. Biodiversity and Conservation 13: Jones, C. G Conservation management of endangered birds. Pages in W. J. Sutherland, editor. Bird ecology and conservation. Oxford University Press, Oxford, United Kingdom. Keo, O., N. J. Collar, and W. J. Sutherland Nest protectors provide a cost-effective means of increasing breeding success in giant ibis Thaumatibis gigantea. Bird Conservation International 19: Kragten, S., J. C. Nagel, and G. R. De Snoo The effectiveness of volunteer nest protection on the nest success of northern lapwings Vanellus vanellus on Dutch arable farms. Ibis 150: Niesten, E., and H. Gjertsen Incentive-based approaches in marine conservation: applications for sea turtles. Conservation and Society 8:5 14. Packman, C. E Seasonal landscape use and conservation of a critically endangered bustard: Bengal florican in Cambodia. Dissertation, University of East Anglia, Norwich, United Kingdom. Pattanayak, S. K., S. Wunder, and P. J. Ferraro Show me the money: do payments supply environmental services in developing countries? Review of Environmental Economics and Policy 4: Prawiradilaga, D. M Ecology and conservation of endangered Javan hawk-eagle Spizaetus bartelsi. Ornithological Science 5: Salvador, D. J. I., and J. C. Ibanez Ecology and conservation of Philippine eagles. Ornithological Science 5: Seng, K. H., B. Pech, C. M. Poole, A. W. Tordoff, P. Davidson, and E. Delattre Directory of Important Bird Areas in Cambodia: key sites for conservation. Department of Forestry and Wildlife, Department of Nature Conservation and Protection, BirdLife International in Indochina and Wildlife Conservation Society Cambodia Program, Phnom Penh, Cambodia. Sok, K., A. Claassen, H. L. Wright, and G. E. Ryan Waterbird nest protection on the Mekong River: a preliminary evaluation, with notes on the recovery and release of white-shouldered ibis Pseudibis davisoni chicks. Cambodian Journal of Natural History 2012: Sommerville, M., J. P. G. Jones, M. Rahajaharison, and E. J. Milner- Gulland The role of fairness and benefit distribution in communitybased Payment for Environmental Services interventions: a case study from Menabe, Madagascar. Ecological Economics 69: Spotila, J. R Sea turtles: a complete guide to their biology, behaviour and conservation. John Hopkins University Press, Baltimore, Maryland, USA. Srikosamatara, S., B. Siripholdej, and V. Suteethorn Wildlife trade in Lao P.D.R. and between Lao P.D.R. and Thailand. Natural History Bulletin of the Siam Society 40:1 47. Sutherland, W. J., A. S. Pullin, P. M. Dolman, and T. M. Knight The need for evidence-based conservation. Trends in Ecology and Evolution 19: Sutrisno, E., Imanuddin, and R. Rachmady Preliminary observations of the breeding biology of the critically-endangered white-shouldered ibis Pseudibis davisoni in East Kalimantan. Kukila 14: Tomillo, P. S., V. S. Saba, R. Piedra, F. V. Paladino, and J. R. Spotila Effects of illegal harvest of eggs on the population decline of leatherback turtles in Las Baulas Marine National Park, Costa Rica. Conservation Biology 22: Tordoff, A. W., R. Timmins, A. Maxwell, K. Huy, V. Lic, and E. H. Khou Biological assessment of the central Indochina dry forest ecoregion. WWF Cambodia Programme, Phnom Penh, Cambodia The Journal of Wildlife Management 77(8)

8 Widmann, P., and I. L. Widmann The cockatoo and the community: ten years of Philippine Cockatoo Conservation Programme. Birding ASIA 10: Wright, H. L., N. J. Collar, I. R. Lake, and P. M. Dolman. 2013a. Amphibian concentrations in desiccating mud may determine whiteshouldered ibis breeding season. Auk 130, in press. Wright, H. L., N. Net, K. Sok, and P. Sum. 2013b. White-shouldered ibis Pseudibis davisoni population size and the impending threat of habitat conversion. Forktail 29, in press. Wright, T. F., C. A. Toft, E. Enkerlin-Hoeflich, J. Gonzalez-Elizondo, M. Albornoz, A. Rodríguez-Ferraro, F. Rojas-Suárez, V. Sanz, A. Trujillo, S. R. Beissinger, V. Berovides Álvarez, X. Gálvez Álvarez, A. T. Brice, K. Joyner, J. Eberhard, J. Gilardi, S. E. Koenig, S. Stoleson, P. Martuscelli, J. M. Meyers, K. Renton, A. M. Rodríguez, A. C. Sosa-Asanza, F. J. Vilella, and J. W. Wiley Nest poaching in neotropical parrots. Conservation Biology 15: Associate Editor: Bruce Thompson. SUPPORTING INFORMATION Additional supporting information may be found in the online version of this article at the publisher s web-site. Figure S1. Daily failure rate (DFR) estimates of whiteshouldered ibis nests in northern Cambodia, , by study site and breeding season: (white); (pale gray); and (dark gray). Data were not available for every breeding season in Lomphat and at the Mekong. We estimated DFR using a binomial logistic regression model (Table S1) and present the number of nests and the number of exposure days (parentheses) above each column; error bars indicate standard errors. Table S1. Parameter estimates for a model of whiteshouldered ibis nest failure in northern and central Cambodia, , across the whole nesting period (incubation to fledging), including effects of study site and breeding season. Kulen Promtep and were reference levels for study site and breeding season respectively. ~AIC c is the change in model Akaike Information Criterion (AIC c ) when the term is removed from the model. Wright et al. Nest Guarding Effectiveness 1617