Communicated by F. Bairlein.

The critically endangered Spoon-billed Sandpiper

The online version contains supplementary material available at

The Spoon-billed Sandpiper

Ten Lincoln–Petersen mark–resighting estimates of the global population size of the Spoon-billed Sandpiper during 2014–2019 indicated a mean population size at the end of the breeding season of 490 mature individuals and 773 individuals of all ages. Two of these estimates were based upon scan surveys at the site of the present study. These estimates also suggested a possible decline at a mean rate of 8% per year (Green et al.

Against this background of suggestive but incomplete evidence of declines in the world population of Spoon-billed Sandpipers, we report here a new analysis of a 9-year series of annual surveys of the species during the boreal winter at Sonadia Island (Chattogram Division, Bangladesh). Zöckler et al. (

Sonadia Island (centre at 21.511° N 91.884° E) is in Maheshkhali subdistrict, Cox’s Bazar district in Chattogram Division, on the coast of southeastern Bangladesh. It covers 49 km^{2} and consists of a variety of coastal wetland habitats, including mudflats, sand dunes, mangroves, sandbars, lagoons, saltpans and beaches (Fig.

Map of the Kaladia mudflats at Sonadia Island, Cox’s Bazar, Bangladesh

We wished to assess the trend in numbers of Spoon-billed Sandpipers using the Kaladia mudflat during a series of nine consecutive boreal winters. The mudflat was used consistently for foraging by Spoon-billed Sandpipers during some parts of the tidal cycle. However, they were not always present in large numbers when mud was exposed and we therefore recognise that counts at this site were unlikely to represent full censuses of the local wintering population because some of the individuals comprising the population probably spent some of their time on other unsurveyed parts of the coast (Chowdhury et al.

In addition to estimating the trend over time in the local population of Spoon-billed Sandpipers, we also assessed changes in the numbers in the same study area of all other shorebird species combined, using the bounded-count method. Our objective was to check whether changes in Spoon-billed Sandpiper numbers were likely to have been caused by changes in habitat, the abundance of invertebrate food or disturbance from human activities, which might be expected to affect the aggregated total of other shorebird species (Zhang et al.

We used standard methods for counting non-breeding shorebirds outlined by Bibby et al. (

We recorded the presence or absence and identity of individual marks on the Spoon-billed Sandpipers in our study area. We did not capture any Spoon-billed Sandpipers for marking within our study area: all of the marked birds recorded had been marked elsewhere. Marks were coloured uPVC leg-flags applied to the right or left tibia and engraved with two black alphanumeric characters (Clark et al.

We conducted scan surveys in November–March of the boreal winters 2017/2018, 2018/2019 and 2019/2020. We located Spoon-billed Sandpipers foraging on the mudflats at Kaladia using binoculars or spotting scopes when the tide level was between 0.5 and 2.5 m relative to the local datum. We aimed to estimate the proportion of Spoon-billed Sandpipers in the local population that had an individually identifiable leg-flag. To do this, we watched each Spoon-billed Sandpiper that we detected to establish whether or not it had a leg-flag. To be eligible for inclusion in the scan survey analysis, each bird had to be observed well enough and at close enough range for the observer to be confident that both tibiae were checked for marks. It was often necessary to watch a bird for several minutes to do this. Resting birds were excluded from the scan sample if they remained sitting and their tibiae could not be seen. Birds that were too distant to be checked or that flew away before a good enough view could be obtained were also excluded. The presence or absence of flags was assessed whenever the observer found an individual which might not have been checked before during that fieldwork session. Birds were not disregarded if the observer realised, by reading the flag, that they had previously recorded the same marked individual recently nearby. Unmarked birds were also recorded when located, despite being suspected to be the same ones as had been recorded previously (Chang et al.

We analysed the monthly count data in two stages. In the first stage, we wished to establish whether the number of birds recorded on the monthly counts showed any consistent tendency to decline during the course of the winter. This was a necessary first step because the bounded-count method assumes that the population is closed within the study period and is only affected by mortality, immigration and emigration to a negligible extent (Robson and Whitlock

We performed an ordinary least squares regression and analysis of covariance with log_{e}-transformed monthly count as the dependent variable. We treated the month of the count (coded 1–5 for November to March) as a continuous covariate and analysed the effect on count of different winters by including it in the model as an eight-level categorical variable. Our objective was to estimate the within-winter trend in log_{e}(count) in relation to survey month and to test whether the regression slopes of log_{e}(count) on survey month varied significantly among winters.

The second stage of the analysis was to estimate the size of the local population of Spoon-billed Sandpipers in each winter from the five monthly counts using the bounded-count method of Robson and Whitlock (_{(k)} and the second-largest count _{(k−1)}. The bounded-count estimate was 2_{(k)} − _{(k−1)} and its 95% confidence interval was: lower bound = _{(k)}, upper bound = _{(k)} + (_{(k)} − _{(k−1)}) 0.95/0.05. We also used the monthly counts of other shorebird species, pooled across all species, to calculate bounded-count estimates of their numbers in each winter by the same method.

We calculated the local population size by dividing a closed-population mark–resighting estimate of the total number of marked individuals present in each winter

We divided the scan survey results for each winter by date into three time periods such that each period had a similar number of scan survey records and had at least ten records. We then used the information on whether each marked individual was recorded or not in each of the three periods to estimate the mean probability of detection ^{3}. We used the asymptotic standard error of

In the second phase of the analysis, we estimated the proportion of Spoon-billed Sandpipers in the local population that were individually marked

Finally, we calculated the Lincoln–Petersen estimate of local population size by dividing the total number of marked individuals detected in each winter

We used ordinary least squares regression to analyse the nine bounded-count estimates and three Lincoln–Petersen estimates. We log_{e}-transformed these values and treated them as the dependent variable in our analyses. We performed the analysis in two stages. We first wished to determine whether the values and trend in relation to year of log_{e}(population) were similar for the bounded-count and Lincoln–Petersen estimates during the 3-year period in which we had both types of estimate. We calculated the difference in log_{e}(population) between the results from the two methods in each winter and obtained the mean and standard error of these differences. We also compared trends estimated by the two methods by regression modelling of log_{e}(population) in relation to year as a continuous covariate, with estimation method coded as a binary variable (bounded-count = 0; Lincoln–Petersen = 1). We used an _{e}(population) values (nine from the bounded-count method and three from Lincoln–Petersen) together in relation to year as a continuous covariate using piece-wise regression, assuming one breakpoint. We determined the breakpoint value (year) iteratively by finding the value at which the residual sums of squares of the fitted model was minimised.

We found no consistent tendency for monthly counts to increase or decrease with time during the course of a winter (Table _{e}(count) on month were negative, the other four being positive. Analysis of covariance indicated that the interaction term between winter (as a factor) and the linear effect of month was non-significant (_{7,27} = 1.855, _{e}(count) and month did not vary significantly among winters. For the model with effects of winter as a factor and with the slope of the within-winter relationship between log_{e}-count and month assumed constant across winters, there was no indication of a marked within-winter trend. The slope of the fitted regression was not significantly different from zero (slope = − 0.0236, _{35} = 0.744,

Monthly counts of Spoon-billed Sandpipers and all other shorebird species combined at Kaladia mudflat, Sonadia Island (Cox’s Bazar, Bangladesh) in nine boreal winters between 2012/2013 and 2020/2021

Boreal winter | Monthly count totals | Largest count | Second largest count | Bounded-count estimate | 95% CI of estimate | ||||
---|---|---|---|---|---|---|---|---|---|

November | December | January | February | March | |||||

2012/2013 | 23 | 16 | 15 | 13 | 11 | 23 | 16 | 30 | 23–156 |

2013/2014 | 20 | 12 | 18 | 19 | 26 | 26 | 20 | 32 | 26–140 |

2014/2015 | 17 | 16 | 23 | 19 | 9 | 23 | 19 | 27 | 23–99 |

2015/2016 | 14 | 17 | 19 | 18 | 21 | 21 | 19 | 23 | 21–59 |

2016/2017 | 17 | 18 | 14 | 15 | 11 | 18 | 17 | 19 | 18–37 |

2017/2018 | 16 | 5 | 16 | 18 | 18 | 18 | 18 | 18 | 18–18 |

2018/2019 | 9 | 10 | 14 | 10 | 10 | 14 | 10 | 18 | 14–90 |

2019/2020 | 5 | 7 | 7 | 6 | 4 | 7 | 7 | 7 | 7–7 |

2020/2021 | 4 | 4 | 3 | 2 | 3 | 4 | 4 | 4 | 4–4 |

2012/2013 | 379 | 594 | 807 | 1406 | 1161 | 1406 | 1161 | 1651 | 1406–6061 |

2013/2014 | 3685 | 2626 | 1224 | 2000 | 941 | 3685 | 2626 | 4744 | 3685–23,806 |

2014/2015 | 884 | 3823 | 1708 | 1585 | 1454 | 3823 | 1708 | 5938 | 3823–44,008 |

2015/2016 | 1006 | 5529 | 977 | 3867 | 1608 | 5529 | 3867 | 7191 | 5529–37,107 |

2016/2017 | 1371 | 4908 | 3444 | 2199 | 1320 | 4908 | 3444 | 6372 | 4908–32,724 |

2017/2018 | 3361 | 1403 | 2680 | 2400 | 2404 | 3361 | 2680 | 4042 | 3361–16,300 |

2018/2019 | 3598 | 7597 | 3088 | 979 | 1638 | 7597 | 3598 | 11,596 | 7597–83,578 |

2019/2020 | 1348 | 2869 | 3801 | 2848 | 509 | 3801 | 2869 | 4733 | 3801–21,509 |

2020/2021 | 1528 | 3907 | 1683 | 1508 | 1205 | 3907 | 1683 | 6131 | 3907–46,163 |

Bounded-count estimates of population size for each winter and 95% confidence intervals (CI) were calculated from the largest and second largest counts

Bounded-count estimates of the local population of Spoon-billed Sandpipers varied among winters between 4 and 32 individuals (Table _{e}(population) between results from the two methods for the same winter was − 0.191 (1 SE = 0.212), indicating that the mean Lincoln–Petersen estimate was 17% lower than the mean bounded-count estimate. However, one of the three Lincoln–Petersen estimates (that for 2017/2018) was 26% higher than the bounded-count estimate and the mean difference was therefore not statistically significantly different from zero (_{2} = 0.900, _{e}(population) in relation to year for the two methods during the 3-year period were broadly similar and both were strongly negative (regression slopes; bounded-count − 0.472 [1 SE = 0.273]; Lincoln–Petersen − 0.763 [1 SE = 0.047]). Analysis of covariance indicated a non-significant interaction term between year and estimation method (_{2,2} = 0.891,

Lincoln–Petersen estimates of the size of the local population of Spoon-billed Sandpipers at Kaladia mudflat, Sonadia Island (Cox’s Bazar, Bangladesh) in three boreal winters between 2017/2018 and 2019/2020

Parameter | Value for this boreal winter | ||
---|---|---|---|

2017/2018 | 2018/2019 | 2019/2020 | |

Number of scan survey dates | 8 | 6 | 7 |

Number of marked individuals recorded for winter | 5 | 2 | 3 |

Estimated number of marked individuals present | 5.5 | 2.1 | 3.0 |

95% CI of | 5.0–8.4 | 2.0–4.0 | 3.0–3.4 |

Number of scan survey records | 127 | 49 | 41 |

Number of scan survey records of marked birds | 31 | 9 | 25 |

Proportion marked | 0.244 | 0.184 | 0.610 |

Clopper–Pearson 95% CI of | 0.172–0.328 | 0.088–0.320 | 0.445–0.758 |

Asymptotic 95% CI of | 0.169–0.319 | 0.075–0.292 | 0.460–0.759 |

Local population estimate | 22.7 | 11.5 | 4.9 |

95% CI of local population estimate | 16.3–38.8 | 6.5–27.7 | 4.0–6.8 |

Considering initially only the bounded-count local population estimates, we found a highly significant tendency for the population to have declined over the 9-year period (Spearman rank correlation _{S} = − 0.967, _{e}(population) (_{2,8} = 14.400,

Bounded-count (filled circles) and Lincoln–Petersen (open circles) estimates of the size of _{e}-transformed values from both methods combined.

During the whole study period we obtained monthly counts of all 25 of the shorebird species (Supplementary information) other than Spoon-billed Sandpiper observed during the study. The bounded-count estimates of the total population of these species combined for each winter showed no clear trend over the nine-year period (Spearman rank correlation _{S} = 0.317,

Our study provides clear evidence of a rapid and accelerating decline in the local population of the critically endangered Spoon-billed Sandpiper during the boreal winters of a nine-year period. It also offers further evidence that Lincoln–Petersen analysis of resightings of individually marked birds is a useful tool to determine local population trends. At the beginning of our study, Sonadia Island was an internationally important wintering area for the species. At that time, it held about 4% of the world population, using the estimate of 773 individuals of all ages for the period 2014–2019 proposed by Green et al. (

From 2017/2018 to 2020/2021, the rate of decline of the Spoon-billed Sandpiper population at Sonadia Island has accelerated and it seems unlikely that it will persist at the site for much longer. Other wintering sites for the species are known in Bangladesh (Bird et al.

Mortality caused by loss of intertidal foraging habitat to land claim has been identified previously as a probable environmental driver of the decline of the global Spoon-billed Sandpiper population (Zöckler et al.

Mortality of Spoon-billed Sandpipers during the non-breeding season caused by hunting has been documented in Myanmar (Zöckler et al.

The rate of decline of the Spoon-billed Sandpiper, based upon repeat surveys at other non-breeding season sites using methods that allow for incomplete detection and Lincoln–Petersen estimates of the world population based upon data from four widely separated monitoring sites (Green et al.

We thank our many donors and supporters for funding our surveys and conservation work on Sonadia Island, Cox’s Bazar, Bangladesh.

Funding acquisition, conceptualization, design, analysis and interpretation: REG and SUC; fieldwork and data collection: all authors; writing: REG and SUC; revision: REG and SUC.

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 18 KB)

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.